Archives for October 2015

The Curious Case of The Disappearing and Reappearing Google Featured Snippet

October 28, 2015 By Glenn Gabe 2 Comments

As Halloween approaches this year, I’m gaining a better understanding of the phrase “Trick or Treat”. Over the past month, I’ve watched Google display a featured snippet for one of my blog posts (the “treat”), only to change that back to a standard snippet (“the trick”). And then back to a featured snippet (“treat”). And you guessed it, back to a standard snippet (“trick”). I really don’t know why that’s happening, but I think it’s an interesting case study.

If you are unfamiliar with featured snippets, then you should read my Search Engine Land column about gaining and losing a featured snippet. In a nutshell, Google can provide an answer at the top of the search results for a query (the featured snippet), along with a link to the third party site containing the content. A featured snippet can also contain visuals, like photos or graphs.

Here’s an example of a featured snippet (since we’ll see plenty of vampires this upcoming Saturday):

Example of a Featured Snippet - Are Vampires Real?

Based on analyzing featured snippets for clients, I know they can drive a massive amount of traffic based on the SERP treatment and the inferred credibility. For example, when Google highlights your content as an answer for a query, separates that content from the rest of the SERP, sometimes provides a visual from your post in the featured snippet, along with a link back to your content. As my case study on SEL documented, you can gain or lose a lot of traffic based on a featured snippet.

It Begins – Pre-processing of a Featured Snippet?
On September 24, 2015, I published my Search Engine Land column explaining how to find queries per url in Google Search Console (GSC). As soon as it was crawled and indexed, I noticed something interesting. It looked like Google was pre-processing a featured snippet.

For example, the description looked like a bulleted list of instructions versus the standard description used for a post. Sure, Google doesn’t always use your meta description, but it also doesn’t always provide a list of items explaining how to do something.

Here is what I saw just 28 minutes after the post was published:

The Pre-processing of a Google Featured Snippet

Needless to say, I wanted to keep an eye on the situation to see if I would receive a featured snippet.

Less Than 2 Weeks Later – It Appears!
I set reminders to check the SERPs daily to see how it was looking. It was about two weeks later that I first saw the featured snippet for a targeted query! And it included a list of six steps from my post along with a visual. It was gorgeous. :)

Featured Snippet With Bulleted List in Google SERPs

So I might have been right! It seems Google was pre-processing the featured snippet by breaking down the steps involved. I remember thinking, “This is AWESOME. I must write about this now.” So I added the topic to my blog post idea bucket and moved on. I was going to try and write the post within a week or so, until…

1 Week Later – Mountain View, We Have a Problem
About a week after noticing the featured snippet, I went to check the SERPs again. I typed in the query, and boom, I saw a standard snippet. That can’t be right… So I opened anther incognito window and checked again. Nope. I opened my chromebook to check on another system. Gone.

Standard Featured Snippet Returns

So it seems Google had already removed the featured snippet! Sure, the SERPs are extremely dynamic, but what could have happened in that one week to remove the featured snippet? The post is a thorough tutorial, it clearly explains how to find queries per url in GSC, and it was published on an authority site (SEL). Come on Google!

So, I took the hit and moved on. But the story doesn’t end there.

2 Weeks Later – I’m Baacckk…
Two weeks later, I was doing some featured snippet detective work for a client and just happened to check the SERP again for my SEL column. And low and behold… the featured snippet was back! But this time it didn’t contain the bulleted list. The visual was still there, but the list was gone. Interesting… but still awesome.

Google Featured Snippet Reappears

Maybe Google realized that the post should yield a featured snippet, but just not in the form of a bulleted list. Or, maybe enough searches for that subject or question occurred to yield a featured snippet. Or maybe this was more testing by Google to see if it warranted a featured snippet at all. Regardless, my featured snippet was back.

The Featured Snippet Returns

3 Days Later – It’s Gone Again
Ugh. Just three days later, the featured snippet was gone again. To be honest, I had no idea what was going on with that featured snippet. I get that featured snippets are algorithmically triggered, so my guess is that the query and/or post is in the gray area for some reason. My post ranks #1 in the SERPs for targeted queries, but does not always yield a featured snippet.

Invisible Man and Google Featured Snippets

Going Straight To The Source: Asking Google
So I decided to go to the source to try and find out what’s going on. I ended up asking both Gary Illyes and John Mueller on Twitter what they thought of the situation. I received an answer, but it was pretty vague.

Gary explained that it’s either ancient aliens (yes, I’m serious) or the fact that Google takes many signals into account when determining when to show a featured snippet. See the Twitter conversation below.

Gary Illyes Regarding Signals That Trigger Featured Snippets:

Gary Illyes Tweet About Featured Snippets

So here we are. Either Google is messing with me or the query/post is in the gray area of featured snippets. Or of course, ancient aliens are involved. :)

Regarding signals that Google takes into account, I started thinking about this more and more after the tweet. What are some of the signals that Google might consider when determining to show a featured snippet? And how could that list help webmasters that are trying to be featured in the SERPs for specific queries?

Based on a quick brainstorm, I have provided some possible signals below. I’m sure this isn’t all of them, but it’s a good start. If you are interesting in driving more featured snippets for your site, then you might want to review this list.

Possible Signals Google Uses When Determining To Show a Featured Snippet:

Type of query (i.e. query formed as a question using what, where, who, why, when).
Query volume for the question at hand.
Content that matches up well with the query (i.e. an answer is clearly provided in an easy to break down format).
Authority of the website that the content is published on.
SERP engagement. Are users interacting with and clicking through the featured snippet? Or are they bypassing that snippet for listings below?
Dwell time while on the destination website. If users bounce back to the SERPs quickly, maybe the content should not yield a featured snippet.
Location of the user. Some featured snippets might be location-specific. i.e. Queries pertaining to a specific country or region.
Device – Will some featured snippets trigger on desktop vs. mobile, or vice versa?
Personalization: Featured snippets based on previous searches and behavior.
Author Rank. Does the authority of the author in that given niche play a factor?

Again, I don’t think this is a final list. If you feel there are others signals to add, let me know in the comments below.

Summary – The Gray Area of Featured Snippets
There you have it. A curious case of a disappearing featured snippet in the Google SERPs. To quickly review, I saw what seemed to be Google pre-processing a featured snippet, followed by receiving the featured snippet with bullets, followed by the featured snippet reverting back to a standard snippet, only to jump back to a featured snippet without bullets, and as of now, it’s standard again.

As Gary Illyes said, there are many factors that determine when a featured snippet should be displayed. I understand that, but in this situation, it seems Google cannot make up its mind. I’ll keep monitoring the situation and update the post if anything changes (and if I learn something new).

Personally, I feel as if my post is in the gray area of featured snippets. It’s either that or some Halloween prank by Google. Maybe RankBrain can help. Muahahaha. :)

Analysis and Findings From The September 2015 Google Algorithm Updates (9/2 and 9/16): Panda 4.2 Tremors, Manual Updates, The Methode Philosophy, and More

October 7, 2015 By Glenn Gabe 8 Comments

September 2015 Google Updates

In my last post, I explained what I have seen during the extended rollout of Panda 4.2. I ended up analyzing over seven weeks of Panda data, since P4.2 is going through an extended rollout. And yes, it’s still rolling out now. More about that soon. Overall, it had been very quiet leading up to that post. Sure, there were some sites that had seen impact and movement, but not like typical Panda updates. It was an underwhelming update to say the least.

Well, something has changed and more significant tremors have been seen across the web.

It’s hard to say if what I’ve been seeing is Panda or a tweak to Google’s core ranking algorithm, but September has been an extremely volatile month algorithm-wise. Specifically, there was movement on 9/2 with some sites seeing large drops in Google organic traffic. And then there was significant movement on 9/16 where some sites surged in traffic, while others dropped heavily.

For example, here’s a site that dropped on 9/16:

Drop During September 16 Google Update

And here’s a site that surged:

Surge During September 16 Google Update

Now, Google pushes hundreds of updates per year, with over one thousand last year according to John Mueller. But the September tremors seem to impact sites with both Panda and Phantom baggage. I’ll cover more about that in this post, but to me it seems like the tremors we have seen are tied to content quality. And again, we have two quality cowboys in town now with Phantom (AKA The Quality Update) and Panda. I definitely saw some sites that got hit hard by Phantom jump during the 9/16 update.

So it’s hard to say it was one over the other. And like I said in my post about Phantom, maybe that was the beginning of Panda being baked into Google’s core ranking algorithm. In other words, they might not be so different…

In this post I’ll cover more about what I’ve seen in September, what Googlers have said about Panda 4.2, I’ll explain more about the connection of these latest updates to Panda and Phantom, and I’ll touch on what I’m calling “The Methode Philosophy”, which is a possible change in how Google views Panda. Let’s jump in.

Confirmed: Panda 4.2 Still Rolling Out & Will For Months
Panda 4.2 started rolling out on 7/18/15 and we knew it would be an extended rollout that could take months. But nobody knew how long that extended rollout would last. Is it still rolling out, has it completed already, and how long before the update completes? All good questions that needed to be answered.

Well, at SMX East Google’s Gary Illyes explained that Panda 4.2 is indeed still rolling out. He last checked on 9/30 and Panda 4.2 was still roaming the web. In addition, he said it will continue to roll out for the next few months. Yes, read that again. Panda 4.2 could have a six month rollout (or longer). As you can imagine, that’s a huge change from previous updates.

Panda 4.2 is still rolling out.

Panda 4.2: Manually Pushed Or Auto-Rollout?
Another question about Panda 4.2 revolves around how it’s being released. Was a giant red button pushed on 7/18 and then P4.2 began its long and extended rollout? Or is Google manually pushing Panda at intervals during the extended rollout. There’s an important distinction between the two.

If Google is pushing the updates manually, then they can refine and tweak the algo based on what they are seeing in the SERPs (to ensure optimal results). That’s similar to what I saw with Panda 4.0 tremors, which was clarified by John Mueller. He said that after a large update, Google can and will tweak that algo to ensure everything is working they way they need it to.

{Updated based on clarification from Gary Illyes:}
So I decided to ask Google’s Gary Illyes on Twitter. At first, Gary explained that Panda 4.2 was being manually pushed at various intervals. But that’s actually not correct. He misunderstood the tweet and didn’t mean that Panda was being pushed manually at various intervals. Instead, Gary explained that Panda 4.2 was manually pushed on 7/18/15 and then automatically rolls out over months. From what he said, there have been no refinements to Panda 4.2 since the initial rollout. That’s important to understand, since it can help us better understand what we are seeing now volatility-wise. And a big thank you to Barry Schwartz for questioning the original response. Without that, I think we would still think Panda 4.2 was being manually pushed.

Both twitter conversations are below. First, here’s my question from yesterday.

Gary Illyes confirms manual Panda 4.2 rollout.
And here’s the clarification from Gary this morning:
Gary Illyes Clarifies Panda 4.2 Rollout

September Volatility – 9/2 and 9/16 To Be Specific
I’m sure you’re wondering what type of volatility I’ve seen in September and how that manifested itself in Google rankings and traffic. Below, I’ll provide some screenshots of what I’ve seen across websites, categories, and countries. I’ll focus on September 16 and highlight any connections I’ve seen with previous Panda and/or Phantom impact.

Google Analytics Trending For A Site Impacted By The 9/16 Update:

GA Drop During Sep 16 Google Update

Google Search Console Trending For A Site Impacted By The 9/16 Update:

GSC Drop in Clicks During 9/16 Google Update

SEMrush Trending For A Site Impacted By The 9/16 Update:

Drop During September 16 Google Update

Google Analytics trending showing a drop during the 9/2 update:

Drop From Sep 2 Google Algorithm Update

Searchmetrics showing a Phantom hit and then positive impact during the 9/16 update:

Phantom and Sep 16 Impact

Another site showing negative impact during Phantom and then an increase during the 9/16 update:

More Phantom and September 16 Movement

SEMrush Trending For A Site Positively Impacted By The 9/16 Update:

Surge In Traffic During Sep 16 Google Update

Searchmetrics Trending For A Site Positively Impacted By The 9/16 Update:

More Upward Movement During 9/16 Update

SEMRush Trending For A Site Seeing Positive and Negative Movement During The 9/2 and 9/16 Updates:

Ups and Downs During September Google Updates

The September Updates Seemed To Target Content Quality (Again)
When analyzing the drop in traffic across sites impacted by the 9/2 and 9/16 updates, I saw a number of familiar problems from a content quality standpoint. These are problems I’ve seen many times while helping companies with Panda and/or Phantom hits. Although some of the sites impacted had historical link problems, not all sites impacted had link issues. But they all had content quality problems.

User Happiness FTW
If you’ve read my previous posts about Panda or Phantom, then you know user happiness is extremely important. If you frustrate or anger users, then you can send them fleeing from your site. During my analysis of the September 2 and September 16 updates, I saw pages that dropped in rankings that provided horrible user experiences. For example, heavy ads scattered throughout the page, forced pagination (for monetization purposes), low quality supplemental content, and more. Let’s explore several of the problems below. Note, if you’ve been following Panda over the years, these problems should not surprise you.

Slideshows or Paginated Content with Keyword Content Not Visible On-Load
I saw several examples of pages that dropped out for keywords when those keywords were not visible on-load. For example, imagine searching for a topic, clicking a result in the search results, landing on a slideshow or paginated article, and not seeing the keywords or topic visible.

This could result in a number of problems SEO-wise. The most important being users getting frustrated and jumping back to the search results since they can’t easily find the content they searched for. Remember, low dwell time is a giant invitation to Panda. Also, John Mueller has said many times that hidden content will either be heavily discounted or not indexed at all. So if one doesn’t get you, the other will. What’s interesting is that pages which used to rank highly for queries related to keywords searched for dropped during the 9/16 update.

September 16 Update Pagination Issues

Thin Pages, Yet Hiding Content
This was a strange one, and it somewhat relates to the bullet above. I found pages that were inherently thin (given the niche), yet the company still chose to force a user to trigger additional content (versus providing it on-load). I have no idea why they would do this on an already thin page, but they did. And I saw those pages drop. It’s worth noting several of those sites also saw drops during the Quality Update (Phantom) in early May of 2015.

September 16 Update and Thin Content

Technical Problems Causing Content Problems (or Ad Problems)
There were a few sites that saw substantial impact during the 9/16 update based on technical problems causing content and/or ad problems. For example, one site built using AngularJS using html snapshots ended up having a giant ad plastered across its content (the snapshot that Google indexes).

Thousands of pages were impacted. On 9/16 they lost approximately 40% of their Google organic search traffic. I can’t say 100% that the drop was due to the ad issue, but it sure looks that way. It’s a leading site in its niche with thorough and detailed content. Although I haven’t analyzed the entire site, having your html snapshots polluted by a giant ad is not good. Not good at all actually.

September 16 Update and Ad Problems

Indexed Bulk Thin Content
A large-scale website with 60M+ pages indexed ended up having over 1M ultra-thin, lower quality pages crawled and indexed (by accident). The business owner already nuked those pages, but they have been on the site for a while. This is also a site that typically sees movement with Panda updates based on its niche, size, and complexity. This is a solid example of “bamboo creep” that can occur without showing any major signs. Then boom, it catches up to you very quickly.

September 16 Update and Bulk Thin Content

Simple, 1997-like Design… INCREASED
I have to explain one jump I saw, since it shocked me at first. The site surged during the 9/16 update, so I dug into the increase in rankings and traffic. When I first visited the site, I almost fell out of my seat. It was so simple, so plain, and so 1997-like, that I couldn’t believe the traffic numbers and strong rankings across competitive keywords.

Simple and Plain Design Surges During September 16 Update

But I picked myself up off the ground and took some time going through the site. It didn’t take long to understand why this site ranked well. It provides granular details for a specific niche, it’s well-documented, and up-to-date. So I reviewed a number of important queries for the site and put myself in the shoes of someone searching for that information. And when I did, I was actually quite happy with the site. It was easy to find the information I wanted, it was organized well (even if it was extremely vanilla), and I found what I wanted quickly.

Remember the first bullet from above? User happiness FTW. :)

Are Bi-Weekly Tremors The New “Quality” Release Schedule?
So, we had an update on Wednesday 9/2 and then another exactly two weeks later on Wednesday 9/16. That sounds a little too perfect, doesn’t it? Based on what we’ve seen, does this mean that Google is tweaking Panda 4.2 and pushing smaller updates every two weeks? Gary says no, so it doesn’t seem like that would be Panda. But it could be Panda 4.2 continuing its rollout. It’s just interesting to see very little impact from P4.2 turn into serious turbulence in early September. And then exactly two weeks later, we saw more turbulence targeting content quality.

That’s speculation, but the updates are similar to the “tremors” I saw after Panda 4.0 (as explained earlier). And by the way, if that bi-weekly schedule is remotely correct, then we should have another tremor any day now. I guess we’ll see.

The Methode Philosophy – A Change In Panda Thinking
Before ending this post, I wanted to bring up an important point based on a video I watched with Google’s Gary Illyes. And I believe it’s a change in Panda philosophy by Google. Specifically, Gary explained that many wrongly believe Panda is about hurting or penalizing a website. He says it’s not. Instead, it’s about ensuring Google “adjusts” websites that become overly prominent for keywords that they shouldn’t rank for. Google wants its users to find the best content possible for the query at hand. So Panda is about making sure websites do not become overly prominent for keywords they shouldn’t rank for. You can watch him explain this in a Q&A with Bruce Clay (at 8:04 in the video).

Now, I totally understand that, and I actually think it’s great. But that’s not how Panda worked in the past. It didn’t just demote pages on an impacted domain for keywords they shouldn’t rank for. Sure, that was part of it, but it crushed websites overall. For example, sites dropped by 60%+ overnight and it wasn’t just overly generic keywords that site shouldn’t rank for. It included core keywords that the site should rank for. That’s why many viewed Panda as an algorithmic penalty.

The reason I bring this up is because Gary’s comments lead me to think that Panda could have changed… and maybe that’s why we are seeing different results from 4.2 than we typically see. Panda 4.2 is supposed to be a refresh. But what if it’s something different? What if it now works based on how Gary explained it?

Again, I like the approach and I think most SEOs would be on board. But I’m concerned for websites that were hit by Panda 4.1 on September 23, 2014 or by the October 24, 2014 update. Many of those websites are still struggling to recover, even after making significant changes. And if Panda has fundamentally changed, then those sites may never recover. They would be in Panda limbo with little hope of coming back, and that would be wrong. It’s hard to say if this is happening, but it’s worth noting.

As The Panda Turns – The Future For Panda and Phantom Victims
After speaking with many webmasters that have been impacted by Panda and/or Phantom, many feel as if they are currently in a horror movie (which fits since Halloween is right around the corner). You know, like being stranded in an abandoned sleep away camp in the woods, knowing a killer could strike at any time, hoping they are spared during the next attack, and trying to send an SOS to local law enforcement. But they just can’t seem to get out of Camp Crystal Lake. Yes, it’s like Friday the 13th, but for SEOs. Just swap Jason for Panda and make it a 6 month movie versus 2 hours. :)

That said, I’m actually hopeful given the volatile September we just experienced. Frequent updates can be good, as more sites can see impact (and hopefully up with the down). Let’s face it, would you rather have 10 months of inactivity or frequent updates? I’m for the latter.

Moving forward, it’s always smart to clean up any problems your site has (both on-site and off-site). This is the same advice I’ve been giving since before Panda 4.2 rolled out and I still believe tackling all quality problems is a smart way to proceed. In a nutshell, you should be ready for more tremors, for Panda 4.3 (if that’s coming), or for the real-time Panda (which we know Google wants to roll out).

That’s all you can do at this point. And of course, you could sip a witches’ brew with a bamboo garnish while trick or treating. But watch out for that Jason Voorhees guy. I heard he’s dressing up as a panda this year. Good luck. :)