On February 19, 2021, we measured a dramatic decrease in recommended snippets on Google SERPs in the United States. Like any responsible data scientist, I waited for it to be a coincidence, did my homework, and published when I was sure I was up to something. Then this happened (30 day view):
C & # 39; MON, GOOGLE! I did all of this nice analysis, found a nice connection between the losses of Featured Snippet, YMYL queries and header terms, and then you go and make me look like a fool ?!
Is there anything we can learn from this strange turn of events? Do I really need this stress Should I just pour myself a cocktail? Don't look forward to any of these answers and more!
Would you like more data? Okay well I think …
Could this recovery be a coincidence of the MozCast 10,000 keyword record? It's unlikely, but let's puncture our i & # 39; s and cross our t & # 39; s. Here is the recommended snippet data from the same time period for approximately 2.2 million US / desktop keywords in the STAT dataset:
So this is going to be a lot more chaotic. We saw a sharp decline on February 19th, followed by a partial rebound, followed by an even larger decline that eventually (for now) landed on a full rebound.
Our original inspection of the waste showed dramatic differences in query length. Here is a breakdown by four word count buckets for prevalence before and after the featured snippet (the data points are February 18, February 19, and March 12):
You can clearly see that the majority of the losses were in one-word queries, with longer queries showing small but far less dramatic decreases. All query lengths were restored through March 12th.
Who really came back from vacation?
If you take two kids on vacation and come back with two kids, you'll be fine, right? What if the kids who came back weren't the same? What if they were robots? Or clones? Or robot clones?
Is it possible that the pages that were tagged with Featured Snippets after the restore were different from those before the deletion? A simple count doesn't tell us the full story even if we slice and dice it. This turns out to be a complicated problem. First, we need to consider that a keyword – in addition to changing the URL of the featured snippet – can win or lose a featured snippet entirely. Consider this pre-drop versus post-recovery comparison:
If you look at the keywords in MozCast that had featured snippets on February 18, 79% of those keywords still had featured snippets on March 12. So we're already down 21%. If we limit the focus to keywords that have retained the recommended snippets and displayed the same page / url in the featured ones, we're reduced to 60% of the original rate.
That seems like a big drop, but we also have to take into account that it was three weeks (22 days to be precise) between the drop and the recovery. How much change is normal in three weeks? For comparison, let's look at the stability of the recommended snippet for the 22 days prior to casting:
While these numbers are slightly better than the numbers after the recovery, we still see three out of ten keywords that either lose a recommended snippet or change the url of the featured snippet. Note that featured snippets are pulled directly from organic first page results, so they keep changing as the algorithm and content of the web evolve.
Do featured snippets stay at home?
It's impossible to tell if Google's original drop was intentional, an unintended consequence of some other (intentional) change, or a mistake. Honestly, given the focus of the drop on so-called "head" inquiries and YMYL (your money, your life) inquiries, I thought this was a conscious change that should stay here. Without knowing why so many Featured Snippets went missing, I can't tell you why they came back, and I can't tell you how long to expect them to stay around.
We can assume that Google will continue to rate the quality of the recommended snippets, especially for queries where result quality is critical (including YMYL queries) or where Google displays knowledge panels and other curated information. Nothing is guaranteed and no tactic is future-proof. We can only keep measuring and adapting.