Tuesday, November 12th, 2013

A Double Penguin Recovery (via 2.1 Update) – But Does It Reveal A Penguin Glitch?

Summary: I analyzed the first double Penguin recovery I have come across during my research (after the Penguin 2.1 update). But what I found could reveal a glitch in the Penguin algorithm. And that glitch could be providing a false sense of security to some business owners.

Double Penguin Recovery is a Double-Edged Sword

If you have followed my blog and Search Engine Watch column, then you know I do a lot of Penguin work.  I started heavily analyzing Penguin 1.0 on April 24, 2012, and have continued to analyze subsequent Penguin updates to learn more about our icy friend.  I’ve had the opportunity to help many companies deal with Penguin hits, and have helped a number recover (and you can read more about those recoveries via several case studies I have written).  It’s been fascinating for sure.  But it just got even more interesting, based on analyzing a site that recovered during Penguin 2.1.   Read on.

Penguin 2.1 rolled out on October 4, 2013, and based on my analysis, it was bigger and badder than Penguin 2.0.   Matt Cutts confirmed that was the case during Pubcon (which was great to hear, since it backed up what I was seeing).  But as I documented in one of my recent Search Engine Watch columns, Penguin 2.1 wasn’t all bad.  There were recoveries, although they often get overshadowed by the carnage.  And one particular recovery during 2.1 caught my attention and deserved further analysis. That’s what I’ll cover in this post.

Ladies and Gentlemen, Introducing The Double Penguin Recovery
I believe it’s important to present the good and the bad when discussing Penguin updates, since there are still some people in the industry who don’t believe you can recover.  But you definitely can recover, so it’s important to document cases where companies bounce back after completing hard Penguin recovery work.

An example of a Penguin recovery:
Example of a Penguin Recovery

Now, there is one thing I hadn’t seen during my past research, and that’s an example of a company recovering twice from Penguin.  I’m not referring to a company that recovers once, gets hit again, and recovers a second time.  Instead, I’m referring to a company that initially recovers from Penguin, only to gain even more during a subsequent Penguin update.

Now that would be an interesting case to discuss… and that’s exactly what I saw during Penguin 2.1.  Interested?  I was too.  :)

Double Penguin Recoveries Can Happen
After Penguin 2.1, I analyzed a website that experienced its second Penguin recovery.  The site was first hit by Penguin 1.0 on April 24, 2012, and recovered in the fall of 2012.  And now, with 2.1 on 10/4/13, the site experienced another surge in impressions and clicks from Google Organic.

The second Penguin recovery on October 4, 2013:
Second Penguin Recovery During 2.1 Update

I’ve done a boatload of Penguin work since 2012, and I have never seen a double Penguin recovery.  So as you can guess, I nearly fell out of my seat when I saw the distinct bump on October 4, 2013.

Penguin Recoveries Lead To Penguin Questions
Based on the second recovery, the big questions for me (and I’m sure you as well), revolve around the reason(s) for the double recovery.  Why did this specific site see another surge from Penguin, when they already did in the past (after hard recovery work)?  Were there any specific factors that could have led to the second recovery?  For example, did they build more natural links, add high quality content, disavow more links, etc?  Or was this just an anomaly?  And most importantly, did Penguin help this website a second time, when it never should have?  In other words, was this a false negative (with the added benefit of a recovery)?  All good questions, and I hope to answer several of them below.

The Lack of Penguin Collateral Damage
I’ve always said that I’ve never seen collateral damage with Penguin.  Every site I’ve analyzed hit by Penguin (now 312), should have been hit.  I have yet to see any false positives.  But with this double recovery, we are talking about another angle with Penguin.  Could a site that shouldn’t see a recovery, actually recover?  And again, this site already recovered during a previous Penguin update.  Could this second recovery be a glitch in Penguin, or were there other factors at play?

History with Penguin
Let’s begin with a quick Penguin history for the website at hand.  It’s an ecommerce website that was devastated by Penguin 1.0 on April 24, 2012.   The site lost close to 80% of its Google Organic traffic overnight.

Initial Penguin Hit on April 24, 2012:
Initial Penguin Hit on April 24, 2012

The site had built thousands of exact match and rich anchor text links over years from spammy directories.  The link profile was riddled with spam.  After the Penguin hit on 4/24/12, their staff worked hard on removing as many links as they could, contacted many directory owners (with some success), and then disavowed what they could not manually remove.  Yes, the disavow tool was extremely helpful for this situation.

The site recovered relatively quickly from Penguin (within two months of finishing the recovery work). The site recovered to about 40% of its original traffic from Google Organic after the Penguin recovery.  That made sense, since the site had lost a majority of links that were once helping it rank for competitive keywords.  Now that the unnatural links were removed, the site would not (and did not) recover to full power.  That’s because it never should have ranked highly for many of the keywords in the first place.  And this is where the site remained until Penguin 2.1.

Initial Penguin recovery in 2012:
Initial Penguin Recovery in 2012

And Along Came Penguin 2.1
After Penguin 2.1 hit, the site experienced an immediate surge in impressions and traffic from Google Organic (and this was crystal clear to see in Google Webmaster Tools).  I’m not sure anyone was expecting a second Penguin recovery, but there it was…  as clear as day.

Impressions were up over 50% and clicks were up close to 60% (when comparing the timeframe after Penguin 2.1 to the timeframe prior).   Checking webmaster tools revealed extremely competitive keywords that were once targeted by Penguin, now gaining in average position, impressions, and clicks.  Certain keywords jumped by 10-15 spots in average position.  Some that were buried in Google were now on page one or page two.  Yes, Penguin 2.1 was providing a second shot in the arm for the site in question.

Impressions and Clicks Increased Greatly After Penguin 2.1 Recovery:
Increase in Impressions and Clicks After Penguin 2.1 Recovery

It was amazing to analyze, but I couldn’t stop several key questions from overpowering my brain.  What changed recently (or over time) that sent the right signals to Google?  Why would the site recover a second time from Penguin?  And could other websites learn from this in order to gain the infamous double Penguin recovery?  I dug into the site to learn more.

What Changed, Why a Second Recovery?
What you’re about to hear may shock you.  It sure shocked me.  Let’s start with what might be logical.  Since Penguin is hyper-focused on links, I reviewed the site’s latest links from across Google Webmaster Tools, Majestic SEO, and Open Site Explorer.

If the site experienced a second Penguin recovery, then I would assume that new links were built (and that they were a heck of a lot better than what got the site initially hit by Penguin).  Google Webmaster Tools revealed a doubling of inbound links as compared to the timeframe when the site first got hammered by Penguin (April 2012).  Majestic SEO and Open Site Explorer did not show as much movement, but did show an increase.

I exported all of the new(er) links and crawled them to double check anchor text, nofollow status, 404s, etc.  And I paid special attention to the links from Google Webmaster Tools, since it showed the greatest number of new links since the first Penguin recovery.  It’s also worth noting that Majestic showed a distinct increase in backlinks in early 2013 (and that includes both the raw number of links being created and the number of referring domains).

Backlinks History Reveals More Unnatural Links Built in Early 2013:
New Unnatural Links Built in Early 2013

Surely the natural, stronger linkbuilding was the reason the site experienced a double Penguin recovery, right?  Not so fast, and I’ll explain more about this next.  It seems Penguin might be glitchy.

More Unnatural Links = Double Penguin Recovery?  Crazy, But True
Believe me, I was really hoping to find stronger, natural links when checking the site’s latest link reporting.  But that wasn’t the case.  I found more spammy links from similar sources that got the site initially hit by Penguin in 2012.  Spammy directories were the core problem then, and they are the core problem now.  Actually, I could barely find any natural links in the new batch I checked.  And that was disturbing.

With all of my Penguin work (having now analyzed 312 websites hit by Penguin), I have yet to come across a false positive (a site that was hit that shouldn’t be hit).  But how about a site recovering that shouldn’t recover?  That’s exactly what this case looks like.  The site built more spammy links after initially recovering from Penguin, only to experience a surge in traffic during Penguin 2.1.  That’s two Penguin recoveries, and again, it’s the first time I have seen this.

The Danger of Heavily Relying on the Disavow Tool
To clarify, I don’t know if the site’s owner or marketing staff meant to build the newer spammy links.  Unnatural links tend to have an uncanny way of replicating across other low-quality sites.  And that’s especially the case with directories and/or article marketing.  So it’s possible that the older, spammy links found their way to other directories.

When You Disavow Links, They Still Remain (and Can Replicate):
The Danger of Relying on the Disavow Tool

This is why I always recommend removing as many links as possible versus relying on the disavow tool for all of them.  If you remove them, they are gone.  If you disavow them, they remain, and can find their way to other spammy sites.

What Does This Tell Us About Penguin?
To be honest, I’m shocked that Penguin was so wrong.  The initial Penguin recovery in 2012 was spot on, as the company worked hard to recover.  They manually removed a significant percentage of unnatural links, and disavowed the rest.  Then they recovered.  But now they experienced a second recovery, but based on the site building more unnatural links (and from very similar sources to the original unnatural links that got them hit in 2012).

So, is this a case of Penguin not having enough data on the new directories yet?  Also, did the company really test the unnatural link waters again by building more spammy links?  As mentioned above, I’ve seen spammy links replicate themselves across low-quality sites before, and that’s especially the case with directories and/or article marketing.  That very well could have happened, although it does look like the links were built during a specific timeframe (early 2013).  It’s hard to say exactly what happened.

Also, will the company eventually get hit by Penguin again (for a second time)?  It’s hard to say, but my guess is the surge in traffic based on Penguin 2.1 will be short-lived.  I cannot believe that the newer, unnatural links will go undetected by our icy friend.  I’m confident the site will get hit again (unless they move quickly now to remove and/or disavow the latest unnatural links).  Unfortunately, the site is teed up to get hit by Penguin.

Summary – Penguin 2.1 Was Wrong (for Now)
This was a fascinating case to analyze.  I have never seen a double Penguin recovery, and I have analyzed hundreds of sites hit by Penguin since April of 2012.  The website’s second recovery looks to be a mistake, as Penguin must have judged the new links as “natural” and “strong”.  But in reality the links were the same old spammy ones that got the site hit from the start.  They were just on different websites.

But as I said earlier, the site is now teed up to get hit by Penguin again. And if that happens, they will lose the power and traffic they have built up since recovering from the first Penguin attack.  If that’s the case, the site will have done a 360 from Penguin attack to Penguin recovery to second Penguin recovery and back to Penguin attack.  And that’s never a good place to be.


Saturday, November 2nd, 2013

How Bing Pre-Renders Webpages in IE11 and How Marketers Can Use The Pre-Render Tag for CRO Today

Bing, IE11, and Pre-rendering Webpages

Bing recently announced it is using IE11’s pre-render tag to enhance the user experience on Bing.com.   Pre-rendering enables Bing to automatically download the webpage for the first search result before you visit that page.  Note, this only happens for “popular searches”, and I’ll cover more about that below.  Pre-rendering via Bing means the destination page will load almost instantaneously when you click through the first search result.  Bing explained that over half of users click the first result, and using IE11’s pre-render tag can enhance the user experience by loading the destination page in the background, after the search is conducted.

A Quick Pre-Render Example:
If I search Bing for “Samsung” in IE11, the first result is the U.S. Samsung website.  When clicking through to the website, the first page loads immediately without any delay (including all webpage assets, like images, scripts, etc.)  Checking the Bing search results page reveals that Bing was using pre-render for the Samsung website homepage.  You can see this via the source code.  See the screenshots below.

Search Results and Sitelinks for Samsung

Checking the source code reveals Bing is pre-rendering the U.S. Samsung homepage:

Bing Source Code Pre-Render Tag


Yes, Google Has Been Doing This With “Instant Pages”
In case you were wondering, Google has been accomplishing this with “Instant Pages” in Chrome since 2011, but it’s good to see Bing roll out pre-rendering as well.  My guess is you’ve experienced the power of pre-rendering without even realizing it.  When Bing and Google have high confidence that a user will click the first search result, they will use the pre-render tag to load the first result page in the background.  Then upon clicking through, the page instantaneously displays.  That means no waiting for large photos or graphics to load, scripts, etc.  The page is just there.

Testing Bing’s Pre-Render in IE11
Once Bing rolled out pre-render via IE11, I began to test it across my systems.  When it kicked in, the results were impressive.  The first result page loaded as soon as I clicked through.  I was off and running on the page immediately.

But when did Bing actually pre-render the page and why did some search results not spark Bing to pre-render content?   Good questions, and I dug into the search results to find some answers.

Identifying Pre-rendering with Bing and IE11
During my testing, I began to notice a trend.  Pre-rendering was only happening when sitelinks were provided for a given search result.  So, if I searched for “apple ipad”, which Bing does not provide sitelinks for, then pre-rendering was not enabled.  But if I searched for just “Apple”, and Bing did provide sitelinks, then pre-render was enabled.  If I searched for “Acura”, sitelinks were provided for the branded search, and the first result was pre-rendered.

A Bing search for “Acura” yields sitelinks:
Search Results and Sitelinks for Acura


Checking the source code reveals Bing is pre-rendering the first search result for “Acura”:
Bing Source Code Pre-Render Tag for Acura


A Bing search for “Derek Jeter” does not yield sitelinks:
Bing Search Results for Derek Jeter
Checking the source code reveals Bing is not pre-rendering the first search result for “Derek Jeter”:
Bing Source Code for Derek Jeter Without Pre-render


So, Bing clearly needed high confidence that I would click through the first listing in order to use pre-render.  In addition, there was a high correlation between sitelinks and the use of the pre-render tag.  For example, “how to change oil” did not yield pre-rendering, “Derek Jeter” did not trigger pre-rendering, and “weather” did not trigger pre-rendering.  But “Firefox” did trigger sitelinks and the use of pre-render.

How Can You Tell If Pre-Rendering is Taking Place
You need an eagle eye like me to know.  Just kidding.  :)  I simply viewed the source code of the search result page to see if the pre-render tag was present.  When it was, you could clearly see the “url0=” parameter and the value (which was the webpage that was being pre-rendered).  You can see this in the screenshots listed above.

And for Chrome, you could check task manager and see if a page is being pre-rendered.  It’s easy to do and will show you if the page is being pre-rendered and the file size.

Using Chrome’s Task Manager to view Pre-rendered Pages
Using Chrome Task Manager to Check Pre-render


How Marketers Can Use Pre-Render On Their Own Websites for CRO Today
Yes, you read that correctly.  You can use pre-render on your own website to pre-load pages when you have high confidence that a user will navigate to that page.  I’m wondering how many Conversion Rate Optimization (CRO) professionals have tried that out!  Talk about speeding up the user experience for prospective customers.

Imagine pre-loading the top product page for a category, the first page of your checkout process, the lead generation form, etc.  Pre-rendering content is supported by Chrome, IE11, and Firefox, so you can actually test this out today.

I’ve run some tests on my own and the pre-rendered pages load in a flash.  But note, Chrome and IE11 support prerender, while Firefox supports prefetch.  That’s important to know if you’re a developer or designer.  Also, I believe you can combine prerender and prefetch in one link tag to support all three browsers, but I need to test it out in order to confirm the combination works.  Regardless, I recommend testing out pre-rendering on your own site and pages to see how it works.

You can analyze visitor paths and determine pages that overwhelmingly lead to other pages.  And when you have high confidence that a first page will lead to a second page, then implement the pre-render tag.  Heck, split test this approach!  Then determine if there was any lift in conversion based on using pre-render to speed up the conversion process.

Analyzing Behavior Flow in Google Analytics to Identify “Connected Pages”:
Analyzing Behavior Flow to Identify Connected Pages


An Example of Using Pre-Render
Let’s say you had a killer landing page that leads to several other pages containing supporting content.  One of those pages includes a number of testimonials from customers, and you notice that a high percentage of users click through to that page from the initial landing page.  Based on what I explained earlier, you want to quicken the load time for that second page by using pre-render.  Your hope is that getting users to that page as quickly as possible can help break down a barrier to conversion, and hopefully lead to more sales.

All that you would need to do is to include the following line of code in the head of the first document:

<link rel=”prerender” href=”http://www.yourdomain.com/some-page-here.htm” >

Note, that will work in Chrome and IE11.  If you combine prerender with prefetch, then I believe that will work across Chrome, IE11, and Firefox.

When users visit the landing page, the second page will load in the background.  When they click the link to visit the page, that page will display instantaneously.  Awesome.


Summary – Pre-Render is Not Just For Search Engines
With the release of IE11, Bing is starting to pre-render pages in the background when it has high confidence you will click the first search result.  And Google has been doing the same with “Instant Pages” since 2011.  Pre-rendering aims to enhance the user experience by displaying pages extremely quickly upon click-through.

But pre-render is not just for search engines.  As I demonstrated above, you can use the technique on your own pages to reduce a barrier to conversion (the speed at which key pages display for users on your website).  You just need to determine which pages users visit most often from other key landing pages, and then implement the pre-render tag.  And you can start today.  Happy pre-rendering.  :)



Monday, October 7th, 2013

Penguin 2.1 Analysis and Findings (Based on the October 4, 2013 Update)

Penguin 2.1 Released on October 4, 2013

On Friday, October 4th at 4:50PM, Matt Cutts announced that Penguin 2.1 was rolling out.  It was shortly after reading that update that I tweeted we could be in for an icy weekend (as the latest update would surely take out more websites).  But on the flip side, a new update also meant that more companies could recover from previous Penguin updates.  Needless to say, I was eager to begin analyzing the impact of our cute, black and white friend.

If you have you followed my blog over the past few years, then you know I do a lot of Penguin work.  So it should be no surprise that I’ve dug into Penguin 2.1 victims to determine new findings, insights, etc.  I fired up my computer at 5:30AM on Saturday morning and started digging in.  Since then, I have analyzed eleven websites (update: now 26 sites) hit by Penguin 2.1 and have provided my findings below.

Note, since this was a minor update (2.1), it signals that the core Penguin algorithm hasn’t been updated since 2.0, but that the data has been refreshed.  Let’s cover what I found during my research.


Fresh Dates, Fresh Spam
While analyzing sites hit by Penguin 2.1, I wanted to check the creation dates for the unnatural links I was coming across.  For most of the sites, the links were first found in late spring 2013, and many were found during the summer.  That makes complete sense, since the sites in question weren’t hit by Penguin 2.0 on May 22.  Instead, they were hit with this latest Penguin update on October 4, 2013.  So, it does seem like fresher unnatural link data is being used.

Penguin 2.1 and Fresh Dates for Unnatural Links


Categories of Unnatural Links Targeted
Since Penguin 2.1 launched, many people have been asking me if the unnatural link footprint has changed since previous Penguin updates.  For example, what types of unnatural links are showing up (for the sites that have been hit by Penguin 2.1).  I have listed my latest findings below (and yes, some familiar unnatural link types showed up):

  • Forum Spam
    I have seen forum spam targeted before, but I saw it heavily targeted with this update.  For example, comments in forum threads that used exact match anchor text links pointing to websites trying to game the system.  During my latest analyses, I saw a lot of forum spam mixed with forum bio spam, which is covered next.
  • Forum Bio Spam
    During my research, I saw forum bios filled with exact match anchor text pointing at sites hit by Penguin 2.1.  This is where the “linkbuilder” set up a profile on a forum, only to use that profile to gain exact match anchor text links to the target website.  I also saw bios targeting multiple websites with exact match anchor text.  This was obviously an attempt to maximize the forum bio to help multiple sites rank.  More about multiple sites soon.
  • Do-Follow Blogs
    I have seen links from various resources that identify do-follow blogs.  A do-follow blog is one that doesn’t add nofollow to links posted (even blog comment signatures in some cases).  The do-follow blog resources are problematic on several levels.  First, they act like a directory using exact match anchor text linking to do-follow blogs.  Second, they absolutely flag certain blogs as being a resource for rich anchor text links (which can send Google down an icy path).Let’s face it, being listed on do-follow resource sites can absolutely send Google a signal that you are trying to game links.  Also, simply finding do-follow blogs and dropping links is not linkbuilding.  If you are doing this, and you got hit by Penguin 2.1, then remove those links as soon as possible.Do-Follow Directories Targeted by Penguin 2.1
  • Blogroll Spam
    OK, an old friend shows up in the list… Just like with Penguin 1.0 and 2.0, spammy blogroll links showed up on sites hit by Penguin 2.1.  This shouldn’t be a shock to anyone involved in SEO, but should reinforce that blogrolls can be extremely problematic when they are on the wrong sites.I believe John Mueller from Google is on record saying that blogrolls overall aren’t bad, but it’s how they are used that can trigger a problem.  I’ve always believed the same thing.  If you have many blogroll links from questionable sites, then I highly recommend attacking them (nuking them, nofollowing them, or disavowing them).  But again, some may be fine.  If you are unsure which ones are bad versus good, ask for help from a seasoned SEO.
  • Spammy Directories
    Another old Penguin friend showed up during my research.  Similar to what I explained above with blogroll spam, certain directories are absolutely Penguin food.  If you have used this tactic in the past, and still have links out there in spammy directories, then nuke them, have them nofollowed, or disavow them.  I’ve seen this category of links show up so many times during my research across Penguin updates, it’s not even funny.  Beware.In addition, I found several sites with millions of inbound links, and many of those were across spammy directories. Let me tell you… if you want to flag your own site, go ahead and build over 2M inbound links from spammy directories.  You’ll get a knock on the door from Mr. Penguin.  That’s for sure.Excessive Links from Spammy Directories
  • Blog Comment Signature Spam
    I came across numerous instances of blog signatures using exact match or rich anchor text.  What’s interesting is that Google seems to target these links, even when they aren’t followed links (most blogs nofollow signature links, and I saw this was the case during my research of sites that were hit by Penguin 2.1).  So, it seems if you were using exact match anchor text as your blog comment signature, then it could be targeted by Penguin (even when those links are nofollowed).
  • (Update) Classified Ads Spam
    As I analyzed more sites hit by Penguin 2.1, I saw low-quality classified websites show up with links pointing at destination sites.  Classified ad listings were used to drop exact match anchor text links, and sometimes in great volume.  For some sites I was analyzing, there were hundreds of pages showing from each domain with links to their websites (from the classified ad websites).  I’ve analyzed many sites hit by Penguin (historically), and haven’t come across many classified websites showing up in the various link profiles.  But with 2.1, I saw this a number of times.


“Linking” Victims Together Via Shared Tactics
One interesting finding I picked up during my analyses was the lumping together of victims.  I noticed forum comment spam and forum bio spam that contained multiple sets of exact match anchor text links (to two different sites).  That even helped me find more Penguin 2.1 victims… as I didn’t know about the second site being hit by Penguin until I found the first one during my research.

So, I’m wondering if Google was able to identify additional targets since they were associated with initial targets.  This would be a brilliant approach for situations where multiple sites were targeted via unnatural links.  It would be a solid example of Google targeting like-minded linkbuilders via evidence it picks up during its crawls.  I can’t say for sure if the other sites would have been targeted anyway by Penguin 2.1, but I saw this numerous times during my latest research of Penguin 2.1.

Linking Targets Together in Forum Spam


Deeper Pages Targeted, P2.0-style
Content-wise, deeper pages were targeted by Penguin 2.1, just like Penguin 2.0.  And since this is a minor update of 2.0, then that makes complete sense.  I’m referring to the target pages of unnatural links on sites hit by Penguin 2.1.   In case you aren’t familiar with what I’m referring to, Penguin 1.0 targeted links to the homepage of a website, where Penguin 2.0 targeted links to any page on the website.

When Matt Cutts first explained this after Penguin 2.0 launched on May 22, it made complete sense to me.  That’s because I had Penguin 1.0 victims ask me why their competitors also weren’t targeted initially by Penguin.  It ends up their competitors had targeted many pages within their own sites versus just driving unnatural links to their homepages.  But in true Google algo fashion, those additional, deeper pages were eventually targeted.  And many sites got hit by Penguin 2.0 (and now 2.1).

Deeper Pages Targeted by Penguin 2.1


How To Recover:
My advice has not changed since Penguin 1.0.  If you have been hit by Penguin, then you need to take the following steps.  And you need to be aggressive with your approach.  If you put band-aids on your Penguin situation, then you might not remove enough links to recover.  And if that’s the case, then you can sit in the gray area of Penguin, never knowing how close you are to recovery.

  • Download Your Links – Download all of your links from multiple sources, including Google Webmaster Tools, Majestic, Open Site Explorer, etc.
  • Analyze Your Links – Analyze and organize your links.  Identify which ones are unnatural and flag them in your spreadsheets for removal.
  • Attack the Links – Form a plan for removing unnatural links.  I always recommend removing as many as you can manually, and then disavowing the remaining links.
  • Removal of Pages is OK – You can also remove pages from your site that are the target pages of spammy links.  John Mueller from Google confirmed that removing pages is the same as removing the links.  Of course, if those are important pages, then you can’t remove them (like your homepage!)
  • Head Down, Keep Driving Forward – Once you have completed your link cleanup, then you’ll need to wait for another Penguin update.  Note, I have seen several Penguin recoveries during Panda updates, so that’s always a possibility.


Summary – Dealing With Penguin 2.1
That’s what I have for now.  I will continue to analyze more sites hit by Penguin 2.1 and will try and write follow-up posts, based on my findings.  If you have been hit, you need to move aggressively to fix the problem.  Be thorough, be aggressive, and move quickly.  That’s the best way to recover from Penguin.  Good luck.



Friday, September 27th, 2013

A Wolf in Panda’s Clothing – How An Expired SSL Certificate Could Impact Organic Search Traffic

Summary: How I helped an ecommerce retailer recover from Panda in eight days, when it was never Panda in the first place.

Expired SSL Certificate Impacting SEO Traffic

A few weeks ago, I had a business owner reach out to me about a potential Panda hit.  His initial email to me was similar to many others I have seen.  He noticed a large drop in Google organic search traffic on a specific date.  And that specific date lined up with an alleged Panda update.  In addition, it was an ecommerce site, which can be susceptible to the wrath of Panda for several reasons.  For example, duplicate content, thin content, technical problems causing perceived content quality issues, etc.

Shortly after receiving the email from the business owner, I also received a call from him.  Yes, he was eager to get moving on recovering from Panda.  It wasn’t long before our call took a left turn, and it was a path this business owner would end up liking.  In addition, it emphasizes an incredibly important point about SEO, Panda, Penguin, and other algorithm updates.  Read on.

It Sure Looks like Panda, But…
Yes, the date of the traffic hit lined up with Panda, but is that enough to clearly say the site was indeed hit by our cute, black and white friend?  Sure, he ran a relatively large ecommerce site, which can open itself up to getting by Panda for several reasons (some of which I mentioned earlier).  But, there are a few points that I firmly believe when it comes to algorithm hits:

1. You need to make sure you know which algorithm update hit your site, or if you were instead hit by a manual action.

2. A thorough SEO audit should be conducted to identify the various problems that could be impacting the site from a Panda, Phantom, or Penguin standpoint.  To me, thorough SEO audits through the lens of an algo update are worth their weight in gold.

So, I asked for the domain again and was planning on performing some checks while we were on the phone.  I wanted to make sure this truly looked like a Panda hit.  One important thing I’ve learned over the years… you never know what you are going to find once you dig in.

Analyzing the Site… Wow, That Was Quick!
I entered the domain name in Chrome and BOOM, I saw the red screen of death.  You know, the one that flags an expired SSL certificate.  It’s scary enough that most people won’t venture any deeper.  I quickly asked the business owner if he was aware of the situation.  He explained that he wasn’t technical, but he knew there was some type of issue with the certificate.  He said his developer had been looking into the problem, but that nothing had changed.

An expired SSL certificate warning in Chrome:
Expired SSL Certificate Warning in Chrome

An expired SSL certificate warning in Firefox:
Expired SSL Certificate Warning in Firefox

So, I checked in Firefox, and BOOM, the same scary message showed up.  I asked if the SSL certificate problem started recently.  He explained that the problem first showed up about the same time he saw the drop in organic search traffic.  So I stopped reviewing the site immediately and explained that we might have just found a “Wolf in Panda’s Clothing”.

The expired SSL certificate was throwing a serious security barrier between his prospective customers and his website.  And the red screen of death is nothing to sneeze at.  The message warning users about the SSL certificate could very well be stopping visitors in their tracks.  And that is what could be impacting the site’s traffic from organic search.  That was my theory anyway.

A Note About SSL Certificates (Along With Expert Information and Advice)
If you run an ecommerce site, then it’s important to understand what an SSL certificate is, and how it can impact your business.  SSL certificates activate the padlock in a web browser when visiting secure sites, and it’s what allows data being sent between the server and the browser to be encrypted.  You can actually view a website’s SSL certificate by right clicking on the padlock and clicking “View Certificate”.  So, if your certificate is expired, the browser is going to warn the user about this.  And that warning could send them running from your website faster than you can say “identify theft”.

Example of an SSL Certificate for JCrew

To explain more about SSL certificates, I asked Brad Kingsley from ORCSWeb for some information.  Brad runs ORCSWeb, which is one of the best hosting providers I have seen in my 18+ years of digital marketing work.  Brad pointed me to Terri Donahue, a Senior Support Specialist and IIS MVP.

Here is a direct quote from Terri:
“An SSL certificate is used to encrypt sensitive data that is being transferred over an insecure network such as the Internet. Without an SSL implementation, when data is transmitted between your server and the recipient of the requested data, it is visible at each ‘hop’ along the way. With an SSL implementation, the data is encrypted until it reaches the destination computer that requested it. This protects the data, such as your credit card number, and ensures that only the requesting entity can decrypt the data for actual viewing.”

“There are a number of trusted Certificate Authorities (CAs) that issue SSL certificates. When purchasing an SSL certificate, these CAs verify the identity of the requestor before issuing the certificate. SSL certificates can be purchased with varying lengths of validity. The shortest term is 1 year with some CAs offering up to 10 years.”

“There are certificate chains that are included in every SSL certificate. The CA has a root or top-level certificate and intermediate certificates that chain to the actual issued SSL certificate that a user purchases. If any of these intermediate chains are not installed on your web server, visitors receive an error when accessing the secured pages of the website. Each vendor is different in the way that these intermediate certificates are obtained. To ensure that all necessary certificate chains are installed, you can check your implementation using this site. This will verify that your certificate is valid and display the full certificate chain.”

SSL Certificate Checker

And here is some important information from Terri regarding SSL certificate expiration (which is exactly what was impacting the business owner I was helping):

“Certificate expiration is handled differently by each CA. Some send notifications to the email address used when the certificate was purchased, while others do not provide any notification. If the CA does not provide notification of expiration, there are multiple ways to handle this. Here is a blog post that refers to a script that can be used to check the expiration of an SSL certificate and send an email when the threshold before expiration is reached. Another way to do this would be to create a database that is maintained with the SSL name and expiration date, which is then monitored, and sends an email at a set period prior to the expiration date.”


Based upon the information Terri provided, you can see that SSL certificates are pretty important.  And since each Certificate Authority (CA) handles expiration differently, expired SSL certificates can throw a wrench into an ecommerce operation pretty quickly.  Now back to our Panda, I mean SSL, problem and how the business owner at hand worked to rectify the situation.

The Quick Fix & The Panda Has Been Exiled
I explained to the business owner that he should address this problem ASAP (like right after we get off the call).  I explained that he should contact me once he renews his SSL certificate, so I could take a quick look to make sure the problem was fixed, and that the red screen of death was gone.  Then we could see if his Google organic search problem turns around.

I received an email from the business owner two days later, and low and behold, his problem was gone.  The mighty Panda had been exiled!  OK, maybe not, but at least the website was humming again with Google organic search traffic.  And that meant revenue was returning to its correct levels.  And yes, this was incredibly important since holiday season was quickly approaching.  The red screen of death would not be good for holiday sales, even if it was red and white like a candy cane. :)

Key Points About SSL Certificates and SEO:
Before I end this post, I wanted to provide some key learnings based on this case study.  If you are an ecommerce retailer, the following bullets could save you from the pain that this business owner felt while his traffic plummeted.

  • Although some business owners aren’t technical, you must understand the ins-and-outs of ecommerce.  That means understanding more about SSL certificates, how they work, when they expire, and other problems that could occur that would inhibit customers from reaching your site.
  • Work with your hosting provider and developer(s) to make sure there are periodic checks of your SSL certificate.  That could nip any serious issues in the bud.  And this includes checking the site across browsers and devices to ensure all is ok.  Avoid the red screen of death at all costs.
  • Keep a master document with information about your SSL certificates, including the certificate authority (CA) that granted the certificate and the expiration date.  Set reminders so you don’t get caught at the last minute (or after the fact).  And you can use the techniques Terri listed above to automate the process of knowing when your certificates will expire.
  • From an SEO standpoint, make sure you know which algorithm update hit your site (or if one hit the site at all).  I worked with this specific business owner to guide his efforts (which led us to the SSL issue without the need for a full-blown Panda audit).  There are times that SEO problems aren’t really SEO problems.  Technical issues that appear at the same time algorithm updates hit can be confusing.  Don’t waste your time, money, and resources tackling the wrong problems.

Summary – Recovering From Panda in 8 Days Can Only Happen If It’s Not Actually Panda
I wish Panda victims could recover in just eight days.  Unfortunately, that’s not the case.  There’s a lot of work that needs to be done to recover from Panda or Phantom.  Luckily for this business owner, Panda didn’t impact his website.  Instead, it was an ecommerce gremlin that attacked his SSL certificate.  And that gremlin was a lot easier to get rid of than a Panda.


Wednesday, September 4th, 2013

Meet The Suggested User List in Google Plus – And How I Jumped From 4K to 80K Followers in Just 4 Months

Suggested User Lists in Google Plus

I’m a big fan of Google+, and have been from the start. I joined the first week it launched, and have been extremely active there ever since.  Since joining, I have loved the engagement, the people, the smart conversations, the functionality, etc.  In addition, I have seen the impact it can have on exposure, SEO, traffic, and credibility.

And like many others who believe in Google+, I have defended it many times.  But, I haven’t defended it based on Google bias.  Instead, I base my opinion and recommendations on data, and my knowledge of how Google works.  To me, if you aren’t using Google+, and you are interested in boosting your social and SEO efforts, then you are missing a huge opportunity.

Growing My Presence on Google+ and Increasing Connections
Similar to what you need to do with any social network, you first need to build a following.  I’ve always believed the best way to do that is to add value and engage users.  And that’s what I tried to do on Google+ from the start.  And over time, I’ve met some incredible people, engaged in some great conversations, while also providing a steady stream of valuable, curated updates.  Those updates included articles, posts, and insights across a range of digital marketing topics like SEO, SEM, Social Media Marketing, Mobile, Web Analytics, etc.

And that approach has definitely worked for me.  I grew my following to about 4K people as of April, 2013.  And no, I really didn’t care about the number of followers.  I cared much more about the quality of people that had me in circles.  I found there were real people on the other end of my updates, with their own experiences, and their own points of view.  I found myself going to Twitter first, but Google+ was always a close second.  It became a natural part of my social media process.

Boom – Hockey Stick Growth
In early April, everything was chugging along as usual on Google+.  I was getting approximately 10-15 followers per day, and that was great.  But then April 12th came along, and I saw first-hand what hockey-stick growth looks like.  My followers began to jump, and jump fast.

I began seeing an increase of 400-600 followers per day.  Yes, you read that correctly.  And on some days, I saw over 1000 new followers.  Since April 12th, I have skyrocketed from 4K followers to over 80K.  You can see the hockey stick growth below in a chart from CircleCount.

Hockey Stick Growth in Google Plus

The Power of the Suggested User List
As you can imagine, I was heavily interested in finding out what was going on.  Was it my latest updates?  Was it my column on Search Engine Watch, combined with my writing elsewhere?  Were the new followers real people?  I had many questions.  So, I started contacting people that had recently circled me asking how they found me.  I figured they could shed some light on why so many people started following since April 12th.

And supporting one of the things I love about Google+ (that there are real people behind the profiles), my new connections got back to me pretty quickly.  The first response basically answered my question: “I saw you on a suggested user list for Technology”.   Then the second came in: “You were on a list suggested by Google”.

Ah, now I got it.  I was added to a suggested user list… now the growth made sense.

Suggested User List in Google Plus for Technology

OK, But Why Was I Added To The SUL?
As I said earlier, I have been continually sharing high quality, curated content on Google Plus (and other social networks).  I determined that my followers would benefit from the specific articles or posts that I read and marked as important.  That could be breaking news, tutorials, reviews of technology, or evergreen digital marketing content.  Never did I use a tool to automate that process.  I simply always tried to keep my connections up to speed on what’s breaking in SEO, SEM, Social, Mobile, Technology, etc.

And over time, Google+ algorithmically determined that the posts I was publishing were high quality, extremely relevant, drove engagement, etc.  At some point, I must have passed a threshold to be added to a suggested user list.  No magic. No gaming.  Just hard work and time.

The Growth of Google+, Seen First-Hand
There has been a lot of talk about Google+ not growing, and that it’s a ghost town.  I never believed that, just based on the people I saw there every day.  And when I was added to the suggested user list, I saw first-hand how many people were signing up.  You see, suggested user lists are presented when you first sign up.  Sure, you can see them any time, but they are prominently showcased when you sign up for Google+.  And if Google+ wasn’t growing, then I wouldn’t have seen a jump to 80K that fast…

What I’ve Changed Since The Jump
Absolutely nothing.
:)  Seriously, why in the world would I change anything now?  I keep doing what I’ve been doing, which is what my overall social strategy has been from the start.  Share valuable posts, engage users, and help others.  And I’ll keep doing that, even if Google+ removes me from the SUL.  It’s how my brain functions, which is also why I do so much blogging.  I love what I do, I love helping other people, and blogging and social media have been a natural fit.

The Impact of Google+ Growth:
OK, I know you’re wondering what the impact has been.  Like everything else in digital marketing, it’s important to look at the impact through a few different lenses.  I’ll touch on each one below.

Engagement and Connections
I have definitely seen an uptick in engagement since the surge began.  That includes people reaching out to me directly, in addition to engagement via posts on Google+.  It’s not 20X engagement, but I’m confident I’m getting in front of more targeted people on a regular basis.  And that’s a good thing social media-wise, and SEO-wise.  More about SEO soon.

Here is an example of a Google Plus Ripple for one of my recent blog posts:

Google Plus Ripple


Clicks From Google+
Based on the total number of updates I share via Google+, Twitter, Facebook, etc., I rarely end up sharing my own links.  My ratio of sharing non-G-Squared posts versus my own content is easily 100 to 1 (if not greater).  I don’t participate in social media to continually shove my own work in front of others.  I share important posts from around the web that relate to digital marketing.  Sure, if I write a post or need to reference one of my own articles, I will.  But that’s not what I do on a regular basis.

In addition, I use bitly pro to have a custom url shortener, which also enables me to track each url.  Using bitly pro, I can see a number of metrics for each url shortened, including total clicks from my shortened link, total clicks from all bitly links for that content, which referrers drove the most clicks on my link, geographic data, etc.  Note, I obviously can’t see impressions, so this is purely click data.

What I’ve seen recently is a jump in the percentage of clicks from Google+.  Now, I’m not Robert Scoble seeing hundreds or thousands of clicks per url.  But, I have seen the percentage climb for G+.  Again, it’s just another signal that more people are seeing my updates and engaging with the content.

For example, over the past 30 days, Google+ accounts for 26% of all clicks on my shortened urls.  Twitter is 24%, Facebook 7%, and “Other” accounts for 35%.  “Other” includes direct traffic, email, air clients, IM, etc.  So, Twitter is actually greater than 24%, based on third party apps.  But, Google Plus is not far behind, accounting for over one quarter of all links I share.

Percentage of Clicks for Google Plus Increases

Authorship and AuthorRank
There’s no doubt that authorship markup has a big impact in the SERPs.  The credibility that author details brings can impact click-through rate for sure.  But in addition to the increased real estate, the pure number of people following me can have an impact.  When users search for answers and solutions, view a search listing from my blog or column with author details, and see 80K+ people have circled me, that can mean a lot towards increased click-through rate.

Author Details in Google SERPs

And from an AuthorRank perspective (which doesn’t impact rankings just yet), the more people that have me in circles could potentially impact my AuthorRank score.  And that score can eventually impact rankings, and subsequent traffic.  Actually, AuthorRank could be impacting in-depth articles already…  So, the more weight my Google+ profile has, the more it could potentially impact AuthorRank (and natural search rankings).  Food for thought.   Note, I cover more about authorship and author stats soon.

With Search Plus Your World (SPYW), Google could increase the rankings of content published by people you are connected with on Google+.  That means my content could surface more often (and higher) for the 80K+ people that are following me.  And that includes both the urls I share from across the web, and the specific Google+ updates I publish.  For example, having plus.google.com/{some-update-here} rank highly in the  SERPs for people that are connected with me on Google Plus.

Here’s a quick example.  I had some connections of mine perform various searches while logged in.  Here is a screenshot of a Google plus update ranking based on a search for the phantom update.  And this plus update does not rank highly when my connection is logged out.

Google Plus Posts Ranking via Search Plus Your World

Author Stats in Google Webmaster Tools
OK, I know after reading that last paragraph about Search Plus Your World, you’re wondering how you can actually see the impact of SPYW via reporting.  For example, is there a way to see how many of your posts show up in the personalized listings?  Google Analytics won’t show you that data, although that would be a great addition.  But, there is a roundabout way to see which plus updates are showing up.  And it’s buried in Google Webmaster Tools in a report titled “Author Stats”.  It can be found under the “Labs” link in the left-side navigation.

Author Stats in Google Webmaster Tools

While researching this post, I dug into my latest stats (which shows statistics for pages which are tied to me as an author).  These are pages that show up in the SERPs with author details (based on rel=author markup).  For me, there are many posts showing up from my Search Engine Watch column, Search Engine Journal Column, and my own blog posts from The Internet Marketing Driver.

But there are other posts listed there…  And as you can guess, they are Google Plus updates.  And those plus updates primarily show for people that are connected to me on Google+ (i.e. people that are seeing my plus updates based on Search Plus Your World when they search for answers).

Over the past 90 days, plus posts account for 15% of the pages showing author details (based on the author stats reporting in Google Webmaster Tools).  That’s a telling story when it comes to Search Plus Your World.  Most of those impressions do not occur for my connections when the personalized results are turned off (or for people that aren’t connected with me on Google+).  So, that’s a clear indicator of the power of SPYW and SEO.  And the more people I’m connected with, the more those plus posts show up in the SERPs.

Google Plus Updates Showing in Author Stats Reporting in Google Webmaster Tools:

Google Plus Posts in Author Stats Reporting in Google Webmaster Tools

And before you go crazy and start screaming that Google is driving more people to its own property (Google+), SPYW also impacts my other updates on Google+, which lead to external websites.  It’s just harder to understand how much they jump up in the SERPs based on Search Plus Your World (since it’s hard to uncover those stats via reporting tools).  The plus posts are easier to isolate, since I can view them in the author stats reporting in webmaster tools.

My Google+ Path Continues
So there you have it.  My story about surging from 4K to 80K followers on Google+ in four months.  Again, employing the right social strategy over the long-term can lead to great things.  You just need to fight through the black hole of social and keep providing value.  Most won’t make it through the black hole, but those that do could see a powerful impact.

My advice to anyone on the fence with Google+ is to start now.  The benefit can be felt on several levels, including increased connections, engagement, exposure, traffic, SEO, and credibility.  But you won’t see any of those benefits if you don’t get started.

I think I can sum up the core point of my post in the following way (and this applies to any social network).  Provide valuable updates, engage users, add value, and help others.  It’s a recipe for hockey stick growth. :)




Monday, August 12th, 2013

Facebook Ads for eCommerce – How To Combine Custom Audiences, Lookalikes, and Unpublished Posts to Target Customers and Similar Users

How to use unpublished posts as Facebook Ads

I used to be extremely critical of Facebook Ads in the past.  But that’s before Facebook released a boatload of functionality for enhancing your campaigns.  Sure, marketplace ads, or ads running the right sidebar, have seen declining engagement over the years, but that’s just a fraction of what you can do now with Facebook Ads.  And I’m finding many advertisers don’t know about the powerful options available to them.

For example, there’s FBX (or retargeting on Facebook), news feed targeting, mobile-only targeting, promoted posts, custom audiences, lookalike audiences, unpublished posts, etc.  And with this enhanced functionality comes better targeting and performance.  Now, I still think paid search can reach someone who is searching for a specific solution at the exact time they need it, and social advertising can’t do that (yet).  But, using advanced targeting within Facebook can absolutely make an impact, and on multiple levels.

In this post, I’m going to explain one method of using three pieces of functionality in Facebook Ads that might change your view of social advertising.  It has for me, and I’ve been using this technique for some time now.  It leverages unpublished posts, custom audiences, and lookalike audiences to target your current customers, and users similar to your customers, when you are running a specific promotion or sale.  It’s a great way to make the most of your current assets, and at a relatively low cost.

Meet Unpublished Posts
I find many business owners have no idea what unpublished posts are.  If you fit into this category, then today is your lucky day.  Unpublished posts enable page owners to create page updates that don’t get shared with their entire fan base.  In addition, you can run ads based on the unpublished posts and use a wealth of ad targeting to reach the right audience (which can include current customers).  Interesting, right?

Unpublished posts in Facebook

The easiest way to create an unpublished post is to use Power Editor.  And if you’re running Facebook Ads and not using Power Editor, you should start today.  It offers a lot of functionality and targeting options not available in Ads Manager (which is what advertisers use on Facebook’s website).

By clicking “Manage Pages” in Power Editor, you can actually craft a page post.  But since we want an unpublished post, you can create the update and not publish it.  That’s ultra-important, since we want to use the post as an ad, and not an update that’s broadcast to your entire fan base.

Creating an unpublished post in Facebook using Power Editor.

So, if you’re an ecommerce provider running a specific sale, you could create an update focusing on that sale, with an understanding it will reach a very specific audience (and not every fan).  I’ll cover how to target specific parts of your customer list soon, including people that are similar to those users.  Once you create your post, you can click your account ID in the left pane to return to your ads dashboard (in Power Editor).

Now we’re ready to talk about custom audiences and lookalikes.

Meet Custom Audiences and Lookalikes
I wrote a post earlier in the year about custom audiences in Facebook.  You should read that post to learn how to set them up.  You’ll need a custom audience in order to use the method I’m covering in this post (since that’s the audience you will target, and it’s also the list you will use to create a lookalike audience).

Custom audiences enable you to upload a list of current customers, based on your in-house email list.  Then, Facebook will match up the list with users on the social network.  Yes, you read that correctly.  That means you can target your in-house email list (or parts of that list) via Facebook Ads.  Awesome, right?

Using Custom Audiences in Facebook

Once your custom audience is created, you can use that list to target current customers with specific promotions and sales.  And you can use unpublished posts to reach them.  Did you catch that?  I said unpublished posts.  That means getting your targeted promotion in front of your current customers (whether they are fans of your page or not).

Great, but what’s a lookalike?
Lookalike audiences enable you to base a new audience (set of Facebook users) on a custom audience (your current customers).  Facebook reviews a number of characteristics about your custom audience (your current customer base), and then finds people similar to your customers.  Yes, once again, eye-opening targeting opportunity ahead.

Imagine you had five custom audiences set up, all containing specific customers for specific categories of products.  Then you could use lookalikes to find similar people (which you can then target via Facebook Ads).  The old days of Facebook ads seem so caveman-like, right?  :)

How To Set Up Lookalikes
Once you have set up a custom audience (following my tutorial), then you can easily select that audience in Power Editor, and choose “Create Similar Audience”.  Choose “Similarity” in the dialog box and Facebook will find users that are similar to your in-house list (based on a number of criteria).  It could take up to 24 hours to create the list, but I’ve seen it take much less time than that (especially for smaller lists).

Using Lookalike Audiences in Facebook

Combining Unpublished Posts, Custom Audiences, and Lookalikes
OK, we have covered unpublished posts that contain targeted messages about new promotions or sales.  We have also covered custom audiences based on our in-house email list.  And, we have covered lookalike audiences, which enable us to target similar people to our own customers.  Now we are ready to tie them together.

1. Create a New Campaign
In Power Editor, you can create a new campaign and set the campaign parameters like name, budget, etc.

Creating a new Facebook campaign in Power Editor.

2. Create a New Ad
Click the “Ads” tab to create your ad.  Under “Type”, choose “Ad”, and then select the radio button labeled “For a Facebook Page Using a Page Post”.  That will enable you to choose an unpublished post for your ad.

Creating an unpublished post ad in Facebook.

3. Choose a Destination
For “Destination”, choose your Facebook Page.  Note, your page’s image and title will still link users to your page, but the post itself can drive users to the sale landing page on your website.  Your post itself is where you should place the link to your landing page (on your own site).  In addition, you should add tracking parameters to your destination urls for your unpublished post (so you can track each campaign via your analytics package).

Choosing an ad destination for unpublished post ad in Facebook.

4. Select An Unpublished Post
Now, choose your unpublished post to use that post as the actual ad.  Note, you can also create your unpublished post at this stage (using Power Editor).  That’s a nice feature that was recently added.

Selecting a page post for an unpublished post ad in Power Editor.

5. Choose your placement:
OK, how awesome is this?  You get to choose where your unpublished post shows up.  For example, in the News Feed (Desktop and Mobile).  This is the most powerful placement in my opinion.  Your ads will show up directly in someone’s news feed versus along the right side.

Choosing ad placement for unpublished post in Power Editor.

6. Choose Your Targeting
Under “Audience”, you can choose targeting, based on the goals of your campaign.  Note, this is not where you will choose your custom or lookalike audience, although the tab is titled “Audience”.  You can choose location, age, gender, etc. if you want more granular targeting than just the custom audiences we created earlier.

Choosing ad targeting for unpublished post in Power Editor.

7. Choose Your Audience (Yes, this is what we’ve been waiting for.)
Under “Advanced Options”, you’ll notice the first field is titled “Custom Audiences”.  If you start typing in that field, your custom audience should show up (based on what you named the audience when you created it).  Once selected, it should show up in the field.  You can leave the rest of the targeting options located below as-is.

Selecting a custom audience for an unpublished post ad in Power Editor.

Clarification Side Note:
To clarify what we’ve been doing, this ad will target your current customer list.  When you create a second campaign, you can choose your lookalike audience.  Then you can run both campaigns and target both your current customer list and people similar to your current customers.   And since they are in separate campaigns, with separate tracking parameters, you can track performance by audience.  Awesome.

8. Select Your Pricing and Status Options
For this example, let’s choose CPC and enter the desired cost per click.  Facebook will provide a suggested CPC to the right.  Once completed, you’re ready to rock.

How to set pricing for an unpublished post ad in Power Editor.

9. Upload Your Campaign
Click “Upload” in Power Editor and your ad will be uploaded to Facebook, where it will need to be approved.  Once approved, you’ll receive a notification that your unpublished post is live.

Uploading an unpublished post ad using Power Editor.

Why this approach works:

1. Exposure and Sharing
By using this approach, you can get your latest sale or promotion in front of your current customers as they browse Facebook, while also providing a great opportunity for that sale or promotion to get shared.  For example a current customer might like your update, and it could hit their friends’ news feeds, which can provide even more exposure and opportunities to land new customers.

2. Engagement
Even though the unpublished post is technically an ad, it still looks and works like a typical page post update.  That means users can like, share, and comment on the post.  And yes, users often do like and comment on unpublished post ads.  Remember, the unpublished post ad is hitting users’ news feeds (both desktop and mobile), so there is a strong chance they will be exposed to your ad.   And if it’s crafted well, then there’s a chance that a certain percentage of that audience will engage with the post. It’s a great way to engage your current customers, while also engaging similar people (via a lookalike audience).

3. Page Likes
Gaining more page likes is an added benefit to using this approach.  Sure, you want people to click through to your sale landing page and buy, but you probably also want more page likes (so you can reach more people with your organic status updates down the line).  I’ve seen unpublished post ads work extremely well for gaining more page likes (across industries).  For example, a recent campaign I launched increased page likes by 7% during a one week period.  Not bad, when you take into account the other benefits from running the campaign (like exposure, sharing, engagement, and sales – which I’ll cover next).

4. Sales (and other important conversions)
Using this approach can yield a low CPA, high ROAS method for increasing sales for specific promotions.  I’ve run campaigns where the CPC was under $0.40 per click, and depending on the specific campaign, return on ad spend (ROAS) can be extremely strong.  For example, 2000 clicks at $0.40 per click is $800.  A conversion rate of 2.0% and an average order value of $75 would yield $3000 in revenue and 275% ROAS.  That’s just a small and quick example, but unpublished page post ads could yield a shot in the arm pretty quickly.

And from a B2B standpoint, with average order values typically much higher than B2C, the ROAS could be even greater.  Even a handful of sales could generated thousands (or tens of thousands) of dollars in revenue.  For example, a recent campaign I launched for a client focused on items starting at $1000 (and some were up to $5000 per item).  Even one sale at $5K based on the campaign I mentioned before would yield a strong ROAS.

And let’s not forget other important micro-conversions on your website.  For example, newsletter signups, which can be a great driver of revenue for any ecommerce provider, app downloads, requests for more information, etc. all fall under this category and can start forging a relationship between prospective customers and your business.

What’s the Downside?
OK, I love using this approach, but social advertising brings some unique challenges with it.  Since what we’ve covered is an actual page post, and not a straight ad, users can interact with it.  That means both positive and negative interaction can occur.  For example, you might have some unhappy customers post their negative feedback in the unpublished page post ad.  How you deal with that situation is for another post, but I always recommend addressing the problem directly (in the post).  But again, there are several situations that can arise, and I’ll try and address them in a future post.  Just keep in mind that users can comment, and those comments might not always be positive.

The Power of Unpublished Posts, Custom Audiences, and Lookalikes
After reading this post, I hope you better understand the power of using unpublished posts along with custom audiences and lookalike audiences.  Unfortunately, the features and functionality I covered in the post are not readily apparent to many Facebook advertisers.  And that’s a shame, since they can be extremely effective for businesses looking to engage current customers and new audiences, while also increasing sales.  I recommend testing this approach soon to see if it can be effective for your business.

You can start today. Create a custom audience, create a lookalike audience, and use Power Editor to create unpublished post ads.  You may never look back.  :)



Monday, July 22nd, 2013

Robots.txt Case Study – How To Block and Destroy SEO with 5K+ Directives

Robots.txt Problems with 5K Directives

If you’ve read my previous posts, then you already know that SEO audits are a core service of mine.  I still believe they are the most powerful deliverable in all of SEO.   So when prospective clients call me for help, I’m quick to start checking some of their stats while we’re on the phone.  As crazy as it sounds, there are times that even a few SEO checks can reveal large SEO problems.   And that’s exactly what happened recently while on the phone with a small business owner who was wondering why SEO wasn’t working for them.

While performing audits, I’ve found some really scary problems.  Problems that spark nightmares for SEOs.  But what I witnessed recently had me falling out of my seat (literally). It was a serious robots.txt issue that was causing serious problems SEO-wise.  I can tell you that I’ve never seen a robots.txt issue like what I saw while on that call.  It was so bad, and so over the top, that I’m adding it as an example to my SEO Bootcamp training (so attendees never implement something like that on their own websites).

Although robots.txt is a simple file that sits at the root of your website, it can still cause serious SEO problems.  The result of the scary-as-heck robots.txt file I mentioned earlier is a small business website with only one page indexed, and all other pages blocked (including its core services pages).  In addition, the robots.txt file was so riddled with problems, that I’m wondering if Googlebot and Bingbot are so offended by the directives that they don’t even crawl the site.  Yes, it was that bad.

Site command revealing only one page indexed:
Site Command Revealing 1 Page Indexed

The SMB SEO Problem
Many small to medium sized businesses are skeptical of SEO companies.  And I don’t blame them.  There are some crazy stories out there, from SEO scams to SEOs not delivering to having all SEO work outsourced to less-experienced third parties.  That leads to many SMBs handling SEO for themselves (which is ok if they know what they are doing).  But in situations where business owner are not familiar with SEO best practices, advanced-level SEO work, etc., they can get themselves in trouble.  And that’s exactly what happened in this situation.

5K Lines of SEO Hell
As you can imagine, the site I mentioned earlier is not performing well in search.  The business owner didn’t know where to begin with SEO and was asking for me assistance (for setting a strong foundation that they could build upon).  That’s a smart move, but I picked up the robots.txt file issue within 5 minutes of being on our call.  I did a quick site: command in Google to see how many pages were indexed, and only one page showed up.  Then I quickly checked the site’s robots.txt file and saw the problems.  And there were really big problems.  Like 5K+ lines of problems.  That’s right, over 5K lines of directives were included in this small business website’s robots.txt file.

Keep in  mind that most small businesses might have five or six lines of directives (at most).  Actually, there are some with just three two or three lines.  Even the most complex websites I’ve worked on had less than 100 directives.  So to see an SMB website with 5K+ lines shocked me to say the least.

Excessive and redundant directives in a robots.txt file:
Example of excessive directives in robots.txt

Yes, the SMB Owner Was Shocked Too
The business owner was not happy to hear what was going on, but still had no idea what I was talking about.  I sent them a link to their robots.txt file so they could see what I was referring to.  I also explained that their web designer or developer should change that immediately.  As of now, it’s still there.  Again, this is one of the core problems when small businesses handle their own SEO.  Serious problems like this can remain (and sometimes for a long time).

What’s Wrong with the File?
So, if you’re familiar with SEO and robots.txt files, you are probably wondering what exactly was included in this file, and why a small business would need 5K+ lines of directives.  Well, they obviously don’t need 5K+ lines, and this was something added by either the CMS they are using or the hosting provider.  It’s hard to tell, since I wasn’t involved when the site was developed.

The file contains a boatload of disallow directives for almost every single directory on the site.  Those directives were replicated for a number of specific bots as well.  Then it ends with a global disallow directive (blocking all engines).  So the file goes to great lengths to disallow every bot from hundreds of directories, but then issues a “disallow all” (which would cover every bot anyway).

Errors in a robots.txt file:
Example of errors in robots.txt

Sitemap Index File Problems
But it gets worse.  The final line is a sitemap directive!  Yes, the file blocks everything, and then tries to feed the engines an xml sitemap that should contain all urls on the site.  But, the site is actually using a sitemap index file, which is typically used to include multiple sitemap files (if you need more than one).  Remember, this is a small business website… so it really shouldn’t need more than one xml sitemap.  When checking the sitemap index file, it only contains one xml sitemap file (which makes no sense)!   If you only have one sitemap file, then why use a sitemap index file??  And that one xml sitemap only contains one URL!!  Again, this underscores the point that the site is creating an overly complex situation for something that should be simple (for most small business websites).

Wrong use of a sitemap index file:
Unnecessary sitemap index file in robots.txt

So let me sum this up for you.  The robots.txt file:

  • Blocks all search engines from crawling all content on the site.
  • Is overly complex by blocking each directory for each bot under the sun.
  • Contains malformed directives (like close to a thousand).
  • Provides autodiscovery for a sitemap index file that only contains one xml sitemap, with only one URL listed!  And all of those urls are blocked by robots.txt anyway!


What To Do –  Get the basics down, then scale.
Based on the audits I’ve performed and the businesses I’ve helped, here is what I think small businesses should do with their robots.txt files:

  • Keep it simple.  Don’t take it from me, listen to Google.  Only add directives when absolutely needed.  If you are unsure what a directive will do, don’t add it.
  • Test your robots.txt file. You can use Google Webmaster Tools to test your robots.txt file to ensure it blocks what you need it to block (and that it doesn’t block what you want crawled).  You can also use some online tools to test your robots.txt file.  You can read another one of my posts to learn about how one of those tools saved a client’s robots.txt file.
  • Add autodiscovery to make sure your clean xml sitemap can be found automatically by the engines. And use a sitemap index file if you are using more than one xml sitemap.
  • And if you’re a small business owner and have no idea what I’m talking about in the previous bullets, have an SEO audit completed.  It is one of the most powerful deliverables in all of SEO and can definitely help you get things in order.


Summary – Avoid 5K Line Robots.txt Files
If there’s one thing you should take away from this post, it’s that the basics are really important SEO-wise. Unfortunately, the small business with the 5K+ line robots.tx file could write blog posts until the cows came home and it wouldn’t matter.  They could have gotten a thousand likes and tweets, and it would have only impacted them in the short term.  That’s because they are blocking every file from being crawled on their website.  Instead of doing that, you should develop a solid foundation SEO-wise and then build upon it.

Nobody needs a 5K line robots.txt file, not even the most complex sites on the web.  Remember, keep it simple and then scale.


Wednesday, July 10th, 2013

Avoiding Dirty Sitemaps – How to Download and Crawl XML Sitemaps Using Screaming Frog

Dirty XML Sitemaps

SEO Audits are a core service I provide, including both comprehensive audits and laser-focused audits tied to algorithm updates.  There are times during those audits that I come across strange pages that are indexed, or I see crawl errors for pages not readily apparent on the site itself.  As part of the investigation, it’s smart to analyze and crawl a website’s xml sitemap(s) to determine if that could be part of the problem.  It’s not uncommon for a sitemap to contain old pages, pages leading to 404s, application errors, redirects, etc.  And you definitely don’t want to submit “dirty sitemaps” to the engines.

What’s a Dirty Sitemap?
A dirty sitemap is an xml sitemap that contains 404s, 302s, 500s, etc.  Note, those are header response codes.  A 200 code is ok, while the others signal various errors or redirects.  Since the engines will retrieve your sitemap and crawl your urls, you definitely don’t want to feed them errors.  Instead, you want your xml sitemaps to contain canonical urls on your site, and urls that resolve with a 200 code.  Duane Forrester from Bing was on record saying that they have very little tolerance for “dirt in a sitemap”.  And Google feels the same way.  Therefore, you should avoid dirty sitemaps so the engines can build trust in that sitemap (versus having the engines encounter 404s, 302s, 500s, etc.)

Indexed to Submitted Ratio
One metric that can help you understand if your xml sitemaps are problematic (or dirty) is the indexed to submitted ratio in Google Webmaster Tools.  When you access the “Sitemaps” section of webmaster tools (under the “Crawl” tab), you will see the number of pages submitted in the sitemap, but also the number indexed.  That ratio should be close(r) to 1:1.  If you see a low indexed to submitted ratio, then that could flag an issue with the urls you are submitting in your sitemap.  For example, if you see 12K pages submitted, but only 6500 indexed.  That’s only 54% of the pages submitted.

Here’s a screenshot of a very low indexed to submitted ratio:
A Low Submitted to Indexed Sitemap Ratio in Google Webmaster Tools

Pages “Linked From” in Google Webmaster Tools
In addition to what I explained above about the indexed to submitted ratio, you might find crawl errors in Google Webmaster Tools for urls that don’t look familiar.  In order to help track down the problematic urls, webmaster tools will show you how it found the urls in question.

If you click the url in the crawl errors report, you will see the error details as the default view.  But, you will also see two additional tabs for “In Sitemaps” and “Linked From”.  These tabs will reveal if the urls are contained in a specific sitemap, and if the urls are being linked to from other files on your site.  This is a great way to hunt down problems, and as you can guess, you might find that the your xml sitemap is causing problems.

Linked From in Google Webmaster Tools
Crawling XML Sitemaps
If you do see a problematic indexed to submitted ratio, what can you do?  Well, the beautiful part about xml sitemaps is that they are public.  As long as you know where they reside, you can download and crawl them using a tool like Screaming Frog.  I’ve written about Screaming Frog in the past, and it’s a phenomenal tool for crawling websites, flagging errors, analyzing optimization, etc.  I highly recommend using it.

Screaming Frog provides functionality for crawling text files (containing a list of urls), but not an xml file (which is the format of xml sitemaps submitted to the engines).  That’s a problem if you simply download the xml file to your computer.  In order to get that sitemap file into a format that can be crawled by Screaming Frog, you’ll need to first import that file into Excel, and then copy the urls to a text file.  Then you can crawl the file.

And that’s exactly what I’m going to show you in this tutorial.  Once you crawl the xml sitemap, you might find a boatload of issues that can be quickly resolved.  And when you are hunting down problems SEO-wise, any problem you can identify and fix quickly is a win.  Let’s begin.

Quick Note: If you control the creation of your xml sitemaps, then you obviously don’t need to download them from the site.  That said, the sitemaps residing on your website are what the engines crawl.  If your CMS is generating your sitemaps on the fly, then it’s valuable to use the exact sitemaps sitting on your servers.  So even though you might have them locally, I would still go through the process of downloading them from your website via the tutorial below.

How To Download and Crawl Your XML Sitemaps

  1. Download the XML Sitemap(s)
    Enter the URL of your xml sitemap, or the sitemap index file.  A sitemap index file contains the urls of all of your xml sitemaps (if you need to use more than one due to sitemap size limitations).  If you are using a sitemap index file, then you will need to download each xml sitemap separately.  Then you can either crawl each one separately or combine the urls into one master text file.  After the sitemap loads in your browser, click “File”, and then “Save As”.  Then save the file to your hard drive.
    Download XML Sitemaps
  2. Import the Sitemap into Excel
    Next, you’ll need to get a straight list of urls to crawl from the sitemap.  In order to do this, I recommend using the “Import XML” functionality in the “Developer” tab in Excel.  Click “Import” and then select the sitemap file you just downloaded.  After clicking the “Import” button after selecting your file, Excel will provide a dialog box about the xml schema.  Just click “OK”.  Then Excel will ask you where to place the data.  Leave the default option and click “OK”.  You should now see a table containing the urls from your xml sitemap.  And yes, you might already see some problems in the list.  :)
    Import XML Sitemaps into Excel
  3. Copy the URLs to a Text File
    I mentioned earlier that Screaming Frog will only crawl text files with a list of urls in them.  In order to achieve this, you should copy all of the urls from column A in your spreadsheet.  Then fire up your text editor of choice (mine is Textpad), and paste the urls.  Make sure you delete the first row, which contains the heading for the column.  Save that file to your computer.
    Copy URLs to a Text File
  4. Unleash the Frog
    Next, we’re ready to crawl the urls in the text file you just created.  Fire up Screaming Frog and click the “Mode” tab.  Select “List”, which enables you to load a text file containing a series of urls.
    List Mode in Screaming Frog
  5. Start the Crawl
    Once you load the urls in Screaming Frog, click “Start” to begin crawling the urls.
    Start a Crawl in Screaming Frog
  6. Analyze the Crawl
    When the crawl is done, you now have a boatload of data about each url listed in the xml sitemap.  The first place I would start is the “Response Codes” tab, which will display the header response codes for each url that was crawled.  You can also use the filter dropdown to isolate 404s, 500s, 302s, etc.  You might be surprised with what you find.
    Analyze a Sitemap Crawl in Screaming Frog
  7. Fix The Problems!
    Once you analyze the crawl, work with your developer or development team to rectify the problems you identified.  The fix sometimes can be handled quickly (in less than a day or two).


Summary – Cleaning Up Dirty Sitemaps
Although XML sitemaps provide an easy way to submit all of your canonical urls to the engines, that ease of setup sometimes leads to serious errors.  If you are seeing strange urls getting indexed, or if you are seeing crawl errors for weird or unfamiliar urls, then you might want to check your own sitemaps to see if they are causing a problem.  Using this tutorial, you can download and crawl your sitemaps quickly, and then flag any errors you find along the way.

Let’s face it, quick and easy wins are sometimes hard to come by in SEO.  But finding xml sitemap errors can be a quick an easy win.  And now you know how to find them.  Happy crawling.




Monday, June 17th, 2013

Panda To Roll Out Monthly Over Ten Days — What It Means for Website Owners and SEOs

Google's New Rollout Plan for Panda

At SMX West, Matt Cutts spoke about the latest Panda changes, and mentioned that the algorithm update had matured.  He explained that Panda will now roll out monthly, but slowly over a 10 day period.  Previously, Panda would roll out every 4-6 weeks and the rollout would be relatively quick (all in one shot).  In addition, Google used to confirm that a Panda update occurred, which meant that massive traffic drops could be tied to Panda.

The upside to the previous approach was that webmasters and SEOs knew when a website was hit by Panda and could begin to analyze their websites through a Panda lens.  Although it’s tough to diagnose exactly why a site was hit by Panda, knowing that the site was indeed hit by Panda was extremely helpful.   So, what does the new Panda rollout mean for webmasters that were hit, or SEOs trying to diagnose the situation?  It means their jobs just got a lot harder.  Let me explain.

Panda Updates in the Past (As Clear As Day)
Let’s revisit what I was explaining about previous Panda updates.  Panda would roll out, webmaster chatter would spike, Barry Schwartz would pick up his bat phone and contact Google, and representatives from Google would confirm the update.  At that point, webmasters knew they were hit by an official Panda update, SEOs knew the site had a quality issues, and they could map out a plan of attack to push back the mighty Panda.

An example of a Panda hit exactly on the day Google confirmed the update:

Panda Drop in Traffic

Fast Forward to Today – The New and Improved Panda
The new Panda will roll out once per month (cool), but will take up to 10 days to fully roll out (uncool).  That means a website could be hit and feel the wrath of Panda at any point during the rollout.  That makes diagnosis much harder.  It’s like going to the doctor with a sore knee 10 days after it happens, but you played basketball, lifted weights, moved furniture, ran a half marathon, got kicked in the knee by your nephew, and fell down the steps during that 10 day period.  Which one of those incidents caused the knee problem?  It’s hard to tell, and that’s my core point.

Google updates its organic search algorithm approximately 500 times per year.  Yes, 500.  In addition, there are other algorithm updates that are pushed out at specific times, like Penguin, Pirate, Above the Fold, the upcoming Smartphone update, etc.  And in addition to that, there are manual penalties where Google specifically takes action on a site that’s violating its webmaster guidelines.  And to make matters even more complex, website owners are making changes all the time.  Redesigns occur, CMS migrations take place, new functionality is added to websites, optimization changes, new pages are added, etc.

Factors That Could Lead to a Drop in Rankings and Traffic

The main point is that isolating a specific date for a massive traffic drop is critically important (and knowing that date matches with a major algo update).   In the past, having a specific date for an update as powerful as Panda was a good thing, while rolling Panda out over 10 days leaves a lot of room (and time) for other factors to be introduced (which could skew your analysis).  And that makes diagnosis tougher, and it also makes recommendations for changes tougher.  The more variables you add to the mix makes the situation exponentially more complex.  And that’s especially the case with Panda, which I’ll cover in greater detail next.

Panda Has Many Tentacles, Penguin Was Acute
Whenever business owners call me about a potential Panda or Penguin hit, I often have to explain the differences between the two.  For many outside of SEO, the two might seem synonymous, when in reality, they are extremely different animals (pun intended).  Penguin is extremely acute (for now).  It hammers unnatural links, and eats them up faster than a good piece of fresh fish.  Panda, on the other hand, targets low-quality content.  And that can mean a lot of different things, depending on the site.

I’ve done  a lot of Panda work since February of 2011, and I’ve seen a number of content-related problems that caused Panda to stomp all over a website.  For example, thin content, duplicate content, affiliate content, cross-linking of domains, low engagement pages, etc. all causing Panda issues.  Even if you knew a website got hit on a certain date, and that was in fact a Panda update, then you still didn’t have a 100% clear understanding of what to actually change.  That led to this post by Google’s Amit Singhal, which listed 23 questions you should ask yourself about your website (if you were wondering what to change after being hit by Panda).  For many, those questions were hard to answer.

23 Questions Panda Victims Should Ask

So, factors leading to a Panda attack were hard to diagnose in the past (since it could be one of many characteristics of a website, or a combination of those).  But now, it gets exponentially harder.  You won’t know 100% that a site was hit by Panda (since it will roll out over a ten day period).  Webmasters will undoubtedly be extremely confused by the drop.  And then SEOs will be tasked with diagnosing the drop, identifying potential Panda problems, and then making tough recommendations for refining the site in question.  Remember, you don’t want to be caught in the grey area of Panda, or you’ll never recover.

Getting Out of the Panda Grey Area (and How The New Rollout Impacts The Exit)
Personally, I’m extremely aggressive with Panda recovery tactics.  There’s a grey area that is very dangerous for business owners.  If you don’t refine enough of your site after getting hit by Panda, then you’ll stay in the grey area.  If you stay in the grey area, and Panda rolls out again, you won’t recover.  And to make matters even tougher, you won’t have any idea how close you were to the light at the end of the Panda tunnel.

That’s madness… which is why I aggressively attack a Panda situation.  I’d rather have a client recover from Panda, and then build upon that foundation, versus having them sit in the grey zone of Panda madness (never knowing how close they are to recovery).

A flowchart I used when helping one client recover from Panda:

Panda Flowchart for Low Quality Content

The latest Panda advancements make the process of attacking Panda potentially harder for many webmasters.  When providing my recommendations for refining a website hit by Panda, I’ve had clients scream out, “are you crazy”, “WTF”, “no freaking way”, “are you sure this is the right route?” etc.  That’s until they calm down, review their analytics reporting again, and their financials.  Once they realize they want/need to recover, they are more open to attacking the problem aggressively.

Well, now that you won’t know 100% that a site has been hit by Panda, then it could make the recovery roadmap harder to get approved and execute.  And that could drag out changes longer than in the past.  And that can mean businesses get caught in the Panda grey area more often.  That could mean that more sites will sit without recovering than in the past.  And believe me, I know how painful it is to be hit by Panda.  I’ve helped a number of companies with Panda attacks.   Just ask a company I helped recover from Panda after six long months of refining the site.

Moving Forward with the New Panda
Now that I’ve explained the situation with the new Panda rollout plan, I’d like to list a few things you can do to deal with the latest Panda advancements.

  • Take a hard look at your website now, and preemptively attack the situation.  Some webmasters know they are doing things wrong… they have a lot of thin content, duplicate content, affiliate content that’s low quality, etc.  I wouldn’t wait for a Panda hit.  I would act now.
  • Analyze and annotate your analytics reporting.  Make sure you know exactly when you were hit, what changes might have occurred recently that could have caused that drop, and use annotations in Google Analytics to document them.  This will only help the SEO that’s going to assist you.
  • If you see a large drop in rankings and Google organic traffic, and you believe it could be due to Panda, then have a technical SEO audit conducted.  I’ve said for a long time that SEO audits provide the most bang for your SEO buck.  A comprehensive audit could reveal many problems associated with your website, including factors that could have led to a Panda attack.
  • Be aggressive with your Panda recovery plan.  If you determine that Panda was the culprit, don’t sit in the grey area like I mentioned earlier.  Be aggressive with your Panda recovery roadmap.  Identify low quality content, gut it, or revamp it.  Be objective, look at engagement numbers, and listen to your SEO (as long as it’s the right SEO).  You might end up removing a good amount of content if you have a serious Panda issue.  That’s ok, but you need to act.


The New Panda – Rolled Out Monthly Over Ten Days Until…
As someone that’s neck deep in SEO, I’m fascinated by algorithm updates.  And that’s especially the case when you watch one (like Panda) move from a manual push to quasi real-time.  That said, I know the confusion this change is going to cause.  Heck, the old Panda caused confusion, and that’s when it rolled out during a specific time, and Google confirmed it.  Now it can hit websites every month, over a ten day period, and Google won’t confirm the update, which means webmasters and SEOs are going to have a heck of a time determining what happened and what needs to change.

I’ll end this post by saying that I don’t think Panda has fully matured yet.  The situation could progress to real-time, where Panda rollouts occur arbitrarily throughout the month (as your site gets recrawled).  And if that’s the case, webmasters and SEOs will be in a Panda warzone.  So I hope you’re stocked up with bamboo. This could get ugly.


Image Source:



Wednesday, May 29th, 2013

Penguin 2.0 Initial Findings – A Deeper Update, But Not Broader [Analysis]

Penguin 2.0 Initial Findings

Penguin 2.0 launched on Wednesday, May 22nd, and it’s an update that most SEOs have been eagerly awaiting.  Leading up to the rollout, all signs pointed to a nasty algorithm update that was going to be bigger and badder than the Penguin 1.0.  There was a lot of speculation about how aggressive it would be, what it would target, and how much impact it would have across the web.  Well, now that it rolled out, was it as nasty as many thought it would be?  Did it target more types of webspam?  And if you are hit, how can you recover?  I’ll try and answer some of these questions below, based on analyzing thirteen sites hit by Penguin 2.0.

In addition to the official Penguin 2.0 update, there was a Phantom Update on May 8th, which could have been Penguin 2.0 being tested in the wild.  I wrote a post explaining my findings, based on analyzing four (now seven) sites hit by that update.  It’s important to read that post as well as this one, so you can start to understand the various factors that led to a drop in rankings and traffic for sites hit on that day.

Analyzing Penguin 2.0
Since Penguin 2.0 launched, I took the same approach that I took with Penguin 1.0.  I’ve been heavily analyzing sites hit by our new, black and white friend.  I’ve been monitoring webmaster forums, receiving emails from P2.0 victims, and digging into each site.  My goal has been to identify consistent factors that impacted sites hit by the latest algorithm update.  I have also been looking for any new additions that Penguin 2.0 might be targeting webspam-wise.

As of today, I have analyzed thirteen sites hit by Penguin 2.0 (I know, unlucky number).  That includes drilling into their link profiles, reviewing their content, and deciphering what led to the Penguin hit.  This post details my initial findings.

Deeper: Yes, Broader: No – Unnatural Links Still the Driving Force, Other Webspam Factors Not Being Targeted
As I explained earlier, we heard that Penguin 2.0 was going to be bigger and nastier than 1.0, but nobody knew exactly what that meant.  Personally, I thought it could mean that more webspam tactics could be targeted, versus just spammy inbound links.  In case you aren’t that familiar with Penguin 1.0, it heavily targeted unnatural links.  If you had a spammy link profile, then you were susceptible to getting pecked.  And a peck could mean a significant drop in Google organic search traffic (sometimes by over 90% overnight).

So, did Penguin 2.0 target additional forms of webspam?  Not in my opinion.  Again, I’ve analyzed thirteen sites hit by P2.0 and all of them had serious link profile issues.  Some had more balanced link profiles than sites hit by Penguin 1.0, but you could easily see the gaming of links from a mile away.  The heavy use of exact match anchor text stood out like a sore thumb.  And some of the sites I analyzed had hundreds of thousands of links using exact match anchor text from questionable sites.  More about those links and websites later in this post.

Penguin 2.0 Heavily Targets Unnatural Links

Homepage vs. Deeper Pages
The one important point to note is the “deeper” reference earlier.  During Twig on 5/22, Matt Cutts announced the release of Penguin 2.0.  During that interview, he explained that Penguin 1.0 only analyzed your homepage links and not pages deeper on your website.  To clarify that point, your domain could still be hit, but Penguin 1.0 only analyzed the homepage link profile to identify the gaming of links.  Looking back, I can see why they launched the first version of Penguin this way.  There were many low quality sites using exact match anchor text leading to their homepages (in an attempt to rank for those keywords).  That’s a good way to launch the first version of Penguin and see how it impacted sites across the web.

But Matt also explained that Penguin 2.0 now analyzed deeper pages on the site.  And that line made a lot of sense to me…  I had some companies reaching out to me after Penguin 1.0 launched complaining that their competitors were using the same tactics they were.  They wanted to know why those companies weren’t getting hit!  Now that we know Penguin 1.0 heavily analyzed the homepage, and didn’t take deeper pages into account, we understand that could be the factor that saved those companies (at least for the time being).  Now that P2.0 is out, those companies using spammy links pointing to deeper pages very well could have gotten hit.

Penguin 2.0 Analyzes Deeper Pages

Am I Seeing The “Deeper Pages” Factor?
I am absolutely seeing websites using exact match anchor text leading to a number of pages on their sites (versus just the homepage).  Actually, every one of the thirteen sites I analyzed had this issue.  So Matt might be telling us the truth when he explained that Penguin 2.0 is deeper.  But again, it’s not broader (taking other webspam tactics into account).

To quickly recap, I have not seen any sign that additional types of webspam were targeted by Penguin 2.0.  It’s still extremely link-based.  I have also seen sites with unnatural links pointing to deeper pages on the site get hit by Penguin 2.0.

Collateral Damage
During major algorithm updates, there are always webmasters that claim they were unfairly hit.  That’s definitely the case sometimes, but I can tell you that I have not seen any collateral damage from Penguin 2.0 first-hand.  All of the sites I have analyzed clearly had unnatural link issues.  And some had extreme unnatural link issues that are going to take a lot of work to rectify… And yes, I can hear the frustration in the voices of the business owners calling me.  Some have a long and tough road ahead of them if they want to recover.

Types of Unnatural Links Remain Consistent
When analyzing unnatural links of sites hit by Penguin 2.0, did the types of unnatural links change at all?  Not from what I can see.  I saw many familiar link types, including comment spam, article sites, spammy directories, blogroll links, link networks (public and private), etc.  Basically, the same types of link manipulation are being targeted by Penguin 2.0 as were targeted by Penguin 1.0 (based on my analysis).

And similar to what I saw with Penguin 1.0, risky sites continually showed up in link profiles.  For example, attack sites, sites hit by malware, etc. I’m not saying that getting hit by malware, or sites that are hacked, get targeted by Penguin 2.0, but a long-term issue without fixing problems like that is a clear signal about the quality of the site.  Think about it, most webmasters hit by malware, or that are being flagged as an attack site, would fix those problems asap.  They wouldn’t let it sit for weeks or months.  I noticed the same situation when analyzing sites hit by Penguin 1.0.

In case you are wondering what a  link scheme is, here is a screenshot from Google Webmaster Guidelines listing various types of link schemes:

Link Schemes

What To Do If You’ve Been Hit
Similar to Penguin 1.0, you need to heavily analyze your link profile to identify unnatural links.  You should organize them by quality and start to create a master list of links to nuke.  And by “nuke”, I don’t mean you should simply disavow all of the unnatural links.  Google wants to know that you tried as hard as possible to manually remove them.  That means setting up a communication plan to webmasters in control of sites that contain spammy links leading to your website.  No, that process isn’t easy, and you can expect a lot of interesting messages back (with some people trying to charge you for link removal).  You can also 404 pages receiving spammy links, but that obviously guts the content on your site.  That’s not the best approach for all situations.

Once you work hard to remove as many links as possible, you can absolutely use the disavow tool for the remaining links.  But again, that shouldn’t be used for the majority of links…  Once you take care of the link situation, you’ll need to wait for another Penguin update in order to see positive movement.  Then again, I have seen Penguin updates during Panda updates (which makes me think they are connected somehow).  You can read my Penguin recovery case studies to learn more about how my clients recovered from Penguin 1.0.

Penguin 2.0 – Now is the time to take action
That’s what I have for now.  I’ll continue analyzing websites hit by Penguin 2.0 and will write follow-up posts covering additional findings.  I’m already helping several clients with dealing with Penguin 2.0, and I anticipate helping more in the coming weeks and months.  If you have any questions, feel free to post them in the comments.  Good luck.