Posted by TheMozTeam
Google let it be known
earlier this year that snippets were a-changin’. And true to
their word, we’ve seen them make two major updates to the feature
— all in an attempt to answer more of your questions.
We first took you on a deep dive of
double featured snippets, and now we’re taking you for a ride
on the carousel snippet. We’ll explore how it behaves in the wild
and which of its snippets you can win.
For your safety, please remain seated and keep your hands, arms,
feet, and legs inside the vehicle at all times!
What a carousel snippet is an how it works
This particular snippet holds the answers to many different
questions and, as the name suggests, employs carousel-like
behaviour in order to surface them all.
When you click one of the “IQ-bubbles” that run along the
initial “parent” snippet with one that answers a brand new
query. This query is a combination of your original search term and
the text of the IQ-bubble.
So, if you searched [savings account rates] and clicked the
“capital one” IQ-bubble, you’d be looking at a snippet for
“savings account rates capital one.” That said, 72.06 percent
of the time, natural language processing will step in here and
produce something more sensible, like “capital one savings
On the new snippet, the IQ-bubbles sit at the top, making room
for the “Search for” link at the bottom. The link is the bubble
snippet’s query and, when clicked, becomes the search query of a
whole new SERP — a bit of fun borrowed from the “People also
You can blame the ludicrous “IQ-bubble” name on Google —
it’s the class tag they gave on HTML SERP. We have heard them
referred to as “refinement” bubbles or “related search”
bubbles, but we don’t like either because we’ve seen them do
both refine and relate. IQ-bubble it is.
There are now 6 times the number of snippets on a SERP
Back in April, we sifted through every SERP in STAT to see just
how large the initial carousel rollout was. Turns out, it made a
decent-sized first impression.
Appearing only in America, we discovered 40,977 desktop and
mobile SERPs with carousel snippets, which makes up a hair over 9
percent of the US-en market. When we peeked again at the beginning
of August, carousel snippets had grown by half but still had yet to
reach non-US markets.
Since one IQ-bubble equals one snippet, we deemed it essential
to count every single bubble we saw. All told, there were a
dizzying 224,508 IQ-bubbles on our SERPs. This means that 41,000
keywords managed to produce over 220,000 extra featured snippets.
We’ll give you a minute to pick your jaw up off the
The lowest and most common number of bubbles we saw on a
carousel snippet was three, and the highest was 10. The average
number of bubbles per carousel snippet was 5.48 — an IQ of five
if you round to the nearest whole bubble (they’re not that
Depending on whether you’re a glass-half-full or a
glass-half-empty kind of person, this either makes for a lot of
opportunity or a lot of competition, right at the top of the
Most bubble-snippet URLs are nowhere else on the SERP
When we’ve looked at “normal” snippets in the past,
we’ve always been able to find the organic results that they’ve
been sourced from. This wasn’t the case with carousel snippets
— we could only find 10.76 percent of IQ-bubble URLs on the
100-result SERP. This left 89.24 percent unaccounted for, which is
a metric heck-tonne of new results to contend with.
Concerned about the potential competitor implications of this,
we decided to take a gander at ownership at the domain level.
Turns out things weren’t so bad. 63.05 percent of bubble
snippets had come from sites that were already competing on the
SERP — Google was just serving more varied content from them. It
does mean, though, that there was a brand new competitor jumping
onto the SERP 36.95 percent of the time. Which isn’t great.
Just remember: these new pages or competitors aren’t there to
answer the original search query. Sometimes you’ll be able to
expand your content in order to tackle those new topics and snag a
bubble snippet, and sometimes they’ll be beyond your reach.
So, when IQ-bubble snippets do bother to source from the same
SERP, what ranks do they prefer? Here we saw another big departure
from what we’re used to.
Normally, 97.88 percent of snippets source from the first page,
and 29.90 percent typically pull from rank three alone. With bubble
snippets, only 36.58 percent of their URLs came from the top 10
ranks. And while the most popular rank position that bubble
snippets pulled from was on the first page (also rank three), just
under five percent of them did this.
We could apply the always helpful “just rank higher” rule
here, but there appears to be plenty of exceptions to it. A top 10
spot just isn’t as essential to landing a bubble snippet as it is
for a regular snippet.
We think this is due to relevancy: Because bubble snippet
queries only relate to the original search term — they’re not
attempting to answer it directly — it makes sense that their
organic URLs wouldn’t rank particularly high on the SERP.
Multi-answer ownership is possible
Next we asked ourselves, can you own more than one answer on a
carousel snippet? And the answer was a resounding: you most
First we discovered that you can own both the parent snippet and
a bubble snippet. We saw this occur on 16.71 percent of our
Then we found that owning multiple bubbles is also a thing that
can happen. Just over half (57.37 percent) of our carousel snippets
had two or more IQ-bubbles that sourced from the same domain. And
as many as 2.62 percent had a domain that owned every bubble
present — and most of those were 10-bubble snippets!
Folks, it’s even possible for a single URL to own more than
one IQ-bubble snippet, and it’s less rare than we’d have
thought — 4.74 percent of bubble snippets in a carousel share a
URL with a neighboring bubble.
This begs the same obvious question that finding two snippets on
the SERP did: Is your content ready to pull multi-snippet duty?
“Search for” links don’t tend to surface the same snippet on the
Since bubble snippets are technically providing answers to
questions different from the original search term, we looked into
what shows up when the bubble query is the keyword being
Specifically, we wanted to see if, when we click the “Search
for” link in a bubble snippet, the subsequent SERP 1) had a
featured snippet and 2) had a featured snippet that matched the
bubble snippet from whence it came.
To do this, we re-tracked our 40,977 SERPs and then tracked
their 224,508 bubble “Search for” terms to ensure everything
was happening at the same time.
The answers to our two pressing questions were thus:
- Strange, but true, even though the bubble query was
snippet-worthy on the first, related SERP, it wasn’t always
snippet-worthy on its own SERP. 18.72 percent of “Search for”
links didn’t produce a featured snippet on the new SERP.
- Stranger still, 78.11 percent of the time, the bubble snippet
and its snippet on the subsequent SERP weren’t a match — Google
surfaced two different answers for the same question. In fact, the
bubble URL only showed up in the top 20 results on the new SERP
31.68 percent of the time.
If we’re being honest, we’re not exactly sure what to make
of all this. If you own the bubble snippet but not the snippet on
the subsequent SERP, you’re clearly on Google’s radar for that
keyword — but does that mean you’re next in line for full
And if the roles are reversed, you own the snippet for the
keyword outright but not when it’s in a bubble, is your snippet
in jeopardy? Let us know what you think!
Paragraph and list formatting reign supreme (still!)
Last, and somewhat least, we took a look at the shape all these
snippets were turning up in.
When it comes to the parent snippet, Heavens to Betsy if we
weren’t surprised. For the first time ever, we saw an almost even
split between paragraph and list formatting. Bubble snippets, on
the other hand, went on to match the trend we’re used to seeing
in regular ol’ snippets: