Central Perk: The Month Google Shook the SERPs

joi, 31 iulie 2014

The Month Google Shook the SERPs

Posted: 30 Jul 2014 05:18 PM PDT

As a group, we SEOs still tend to focus most of our attention on just one place – traditional, organic results. In the past two years, I've spent a lot of time studying these results and how they change over time. The more I experience the reality of SERPs in the wild, though, the more I've become interested in situations like this one (a search for "diabetes symptoms")...

See the single blue link and half-snippet on the bottom-left? That's the only thing about this above-the-fold page that most SEOs in 2014 would call "organic". Of course, it's easy to find fringe cases, but the deeper I dig into the feature landscape that surrounds and fundamentally alters SERPs, the more I find that the exceptions are inching gradually closer to the rule.

Monday, July 28th was my 44th birthday, and I think Google must have decided to celebrate by giving me extra work (hooray for job security?). In the month between June 28th and July 28th, there were four major shake-ups to the SERPs, all of them happening beyond traditional, organic results. This post is a recap of our data on each of those shake-ups.

Authorship photos disappear (June 28)

On June 25th, Google's John Mueller made a surprise announcement via Google+:

We had seen authorship shake-ups in the past, but the largest recent drop had measured around 15%. It was clear that Google was rethinking the prevalence of author photos and their impact on perceived quality, but most of us assumed this would be a process of small tweaks. Given Google's push toward Google+ and its inherent tie-in with authorship, not a single SEO I know had predicted a complete loss of authorship photos.

Yet, over the next few days, culminating on the morning of June 28th, a total loss of authorship photos is exactly what happened:

While some authorship photos still appeared in personalized results, the profile photos completely disappeared from general results, after previously being present on about 21% of the SERPs that MozCast tracks. It's important to note that the concept of authorship remains, and author bylines are still being shown (we track that at about 24%, as of this writing), but the overall visual impact was dramatic for many SERPs.

In-depth gets deeper (July 2nd)

Most SEOs still don't pay much attention to Google's "In-depth Articles," but they've been slowly gain SERP share. When we first started tracking them, they popped up on about 3.5% of the searches MozCast covers. This data seems to only get updated periodically, and the number had grown to roughly 6.0% by the end of June 2014. On the morning of July 2nd, I (and, seemingly, everyone else), missed a major change:

Overnight, the presence of in-depth articles jumped from 6.0% to 12.7%, more than doubling (a +112% increase, to be precise). Some examples of queries that gained in-depth articles include:

xbox 360
hotels
raspberry pi
samsung galaxy tab
job search
pilates
payday loans
apartments
car sales
web design

Here's an example set of in-depth for a term SEOs know all too well, "payday loans":

The motivation for this change is unclear, and it comes even as Google continues to test designs with pared down in-depth results (almost all of their tests seem to take up less space than the current design). Doubling this feature hardly indicates a lack of confidence, though, and many competitive terms are now showing in-depth results.

Video looks more like radio (July 16th)

Just a couple of weeks after the authorship drop, we saw a smaller but still significant shake-up in video results, with about 28% of results MozCast tracks losing video thumbnails:

As you can see, the presence of thumbnails does vary day-to-day, but the two plateaus, before and after June 16th, are clear here. At this point, the new number seems to be holding.

Since our data doesn't connect the video thumbnails to specific results, it's tough to say if this change indicates a removal of thumbnails or a drop in rankings for video results overall. Considering how smaller drops in authorship signaled a much larger change down the road, I think this shift deserves more attention. It could be that Google is generally questioning the value and prevalence of rich snippets, especially when quality concerns come into play.

I originally hypothesized that this might not be a true loss, but could be a sign that some video snippets were switching to the new "mega-video" format (or video answer box, if you prefer). This does not appear to be the case, as the larger video format is still fairly uncommon, and the numbers don't match up.

For reference, here's a mega-video format (for the query "bartender"):

Mega-videos are appearing on such seemingly generic queries as "partition", "headlights", and "california king bed". If you have the budget and really want to dominate the SERPs, try writing a pop song.

Pigeons attack local results (July 24th)

By now, many of you have heard of Google's "Pigeon" update. The Pigeon update hit local SERPs hard and seems to have dramatically changed how Google determines and uses a searcher's location. Local search is more than an algorithmic layer, though – it's also a feature set. When Pigeon hit, we saw a sharp decline in local "pack" results (the groups of 2-7 pinned local results):

We initially reported that pack results dropped more than 60% after the Pigeon update. We now are convinced that this was a mistake (indicated by the "?" zone) – essentially, Pigeon changed localization so much that it broke the method we were using. We've found a new method that seems to match manually setting your location, and the numbers for July 29-30 are, to the best of my knowledge, accurate.

According to these new numbers, local pack results have fallen 23.4% (in our data set) after the Pigeon update. This is the exact same number Darren Shaw of WhiteSpark found, using a completely different data set and methodology. The perfect match between those two numbers is probably a bit of luck, but they suggest that we're at least on the right track. While I over-reported the initial drop, and I apologize for any confusion that may have caused, the corrected reality still shows a substantial change in pack results.

It's important to note that this 23.4% drop is a net change – among queries, there were both losers and winners. Here are 10 searches that lost pack results (and have been manually verified):

jobs
cars for sale
apartments
cruises
train tickets
sofa
wheels
liposuction
social security card
motorcycle helmets

A couple of important notes – first, some searches that lost packs only lost packs in certain regions. Second, Pigeon is a very recent update and may still be rolling out or being tweaked. This is only the state of the data as we know it today.

Here are 10 searches that gained pack results (in our data set):

skechers
mortgage
apartments for rent
web designer
long john silvers
lamps
mystic
make a wish foundation
va hospital
internet service

The search for "mystic" is an interesting example – no matter what your location (if you're in the US), Google is showing a pack result for Mystic, CT. This pattern seems to be popping up across the Pigeon update. For example, a search for "California Pizza Kitchen" automatically targets California, regardless of your location (h/t Tony Verre), and a search for "Buffalo Wild Wings" sends you to Buffalo, NY (h/t Andrew Mitschke).

Of course, local search is complex, and it seems like Google is trying to do a lot in one update. The simple fact that a search for "apartments" lost pack results in our data, while "apartments for rent" gained them, shows that the Pigeon update isn't based on a few simplistic rules.

Some local SEOs have commented that Pigeon seemed to increase the number of smaller packs (2-3 results). Looking at the data for pack size before and after Pigeon, this is what we're seeing:

Both before and after Pigeon, there are no 1-packs, and 4-, 5-, and 6-packs are relatively rare. After Pigeon, the distribution of 2-packs is similar, but there is a notable jump in 3-packs and a corresponding decrease in 7-packs. The total number of 3-packs actually increased after the Pigeon update. While our data set (once we restrict it to just searches with pack results) is fairly small, this data does seem to match the observations of local SEOs.

Sleep with one eye open

Ok, maybe that's a bit melodramatic. All of the changes do go to show, though, that, if you're laser-focused on ranking alone, you may be missing a lot. We as SEOs not only need to look beyond our own tunnel vision, we need to start paying more attention to post-ranking data, like CTR and search traffic. SERPs are getting richer and more dynamic, and Google can change the rules overnight.

Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don't have time to hunt down but want to read!