I'm a passionate, and focused entrepreneur, leading a very talented team of people at STAT Search Analytics. By unlocking insight in complex search engine data, we help SEO teams achieve amazing results.
We're based out of Railtown, in Vancouver. We're an unusual company, being completely private, bootstrapped, and profitable. We've experienced high growth over the past years and the team is now 45 people strong.
We are currently looking for:
- Full Stack Developers
- Data Engineers
- DevOps Engineers
What you will learn?
Featured Snippets: From Then to Now, Volatility, and Voice Search
Thanks, Rob. Hi everyone. I’m Rob. Welcome to Rob Fest here at SearchLove. Dana, okay. So I’m from Canada too, Distilled loves Canadians apparently.
I want to talk about Trader Joe’s for just a quick second. Okay, so I’m obsessed as well. We’re so obsessed with Trader Joe’s in Canada that there’s this dude in Vancouver who like drives down to Seattle, buys a whole bunch of stuff in bulk from Trader Joe’s and resells it in Vancouver and the store’s called Pirate Joe’s.
And Trader Joe’s sued him and so he changed the name of the store to Irate Joe’s, he just dropped the P, which is like so good. He’s out of business now, but I applaud him wherever he is because man, that’s amazing.
This is a weird question. But this is like a show of hands, who hasn’t heard of STAT before?
It’s like incredible. I’m not going to bother with the like elevator pitch nonsense because that’s incredible, the amount of people who recognize who STAT is or who we are. I’ll instead use my time here to talk a little bit about what I’m grateful for.
I have a really great team back in Vancouver who help us lift an incredible amount of data. We have access to just massive reams of data. And in order to make it sing for research, like what I’m going to present today, it takes a lot of people. It’s not just me and so they’re not here all of them, but I want to give a big shout out to my team back in Vancouver for all their hard work. I also just want to take a moment and express my gratitude to the Distilled team for the opportunity to be here. It’s just like phenomenal that I have an opportunity to come up on the stage and share my passion with you also. Thank you very much to the team from Distilled for putting on a great show.
I’m going to talk about featured snippets. Surprise! They totally took my thunder away from me. But here’s what we’re going to talk about today, okay?
You guys probably think about me or my company as the featured snippet folks and I can’t blame you. We talked about them a lot. We did the study about like two years ago and ever since then people keep wanting us to talk about featured snippets. And every time I say the word featured snippets, I start to cringe because I’ve said it so many times in my life, but I’m excited to be here and give you an update because there are some interesting things happening right now. We’re seeing digital assistants get more popular than ever, Google’s digital assistant and their voice search efforts are heavily reliant on featured snippets. So it’s super timely and we’re going to walk through my favorite SERP feature and it just so happens that Danny Sullivan is Google’s public liaison for search, like who saw that happening?
He wrote this article called A Reintroduction to Google Speech and Snippets and in it he cites that the traditional 10 Blue Links paradigm no longer flies in today’s world. So we were like working on this topic back home and we saw this post go up and we were like, yeah, we’re Google-approved. So we’re super excited to go through this data today.
Here’s what I’m going to do. We’ve been tracking the data for two years. We’ve done frequent check-ins on the data and we’re going to use that information to look at like the history, the evolution and where they’re at today. Then we’re going to move into some discussion around the viability of a snippet strategy in the long-term, like is it a safe investment? And once you win it, how hard do you have to keep it? Sorry how hard you have to fight to keep it? And that’s called snippet volatility. Then we’re going to focus on the cozy relationship between snippets and voice search. It’s super cozy.
If you have any questions as I go through, you can hit me at Twitter @STATrob. You get my company, I get STAT, put all the criticisms to my company not to me and if you’re like really geeky for this stuff and you want to get deep into it, go to getstat.com/SearchLove, everything I’m talking about, all the slides, the white papers, the data. It’s all available there for you to get super super deep into it.
Alright. Let’s do this.
The Evolution of Featured Snippets
Okay. So in order for us to accurately track the progression of featured snippets, we need to use a consistent keyword set. Right? So we have 1 million keywords, they’re high in CPC terms and we’ve been tracking them since January 2016. That consistent set gives us a consistent sample and signal to measure with and whenever we get the data into STAT, the first thing we want to do is say, well, “How many snippets are there today?”
Sorry heads up getting rid of that. There you go.
We want to say how many steps are there today and it turns out there’s a lot. In our last run, 31% of the keywords in our 1 million keyword set return to featured snippet. There’s like a 230% increase over time since January 2016. So if you don’t want to take my word or Danny Sullivan’s word for it that snippets are more important than ever, there’s your proof.
Featured Snippet Rank
The next thing you want to know is how visible are they because that visibility is perhaps the coolest part about featured snippets, right? You’re at the very top of the SERP and Google is like taking your content, put a big box around it and said this. Like this is the best content I can give you to answer this particular query of yours. So the authority that bestows upon you and your content on your site is massive and we want to know, is that still the case? Is it still at the top of the SERP?
I don’t know why I point with this thing. It’s like I don’t need to point.
Anyway, so it turns out it’s not. Actually 4% of the time, snippets are now in the second place and we saw that and we’re like whoa, like wait a minute. We always used to expect snippets to be at the very top and what could possibly be pushing them down 4%, not a huge drop, but it’s interesting.
It turns out it’s shopping boxes. We also see Google News boxes and we see some home service ads as well. But it’s interesting to note that Google is willing to rank things above the snippet and they’re displaying that they have a willingness to do this for certain areas that are important to them. The effect this has whether good or bad on the traffic that a snippet can send is hard for us to measure but it’s interesting to note that this is a possibility and if you’re in a space where there are either like e-commerce terms or current event terms, you might want to be aware that you could be outranked by another box above your snippet.
Featured Snippet Source Position
The next thing that’s super compelling about snippets is what we call source position. The idea here is that you can rank in like position five and you can leapfrog on all the competitors and also appear on the snippet. So you get these opportunities for like great double exposure on the SERP and this term, the traditional paradigm of like just ranked higher kind of on its head because people realize that they didn’t have to be number one to be number one.
When we first looked at this we saw like around 30% of the time, snippets were sourced from position number 2 and then there was this really nice diminishing curve of source position. And we wanted to know is this still the case? Interestingly, it’s not.
The curve looks the same right? But from January until November over a year later, we’ve seen that shift down by one point and we might look at that and think, you know cool like that means the bar is lower. We don’t have to rank as high to be eligible for a feature snippet. Unfortunately, that’s not the case. Because most of the time the people also ask box is the second thing on the SERP. So the most common sequence of results that we see now that’s different than January 2016 is featured snippet, PAA, organic result. And around 31% of the time, that organic result is the source of the featured snippet.
This kind of gets back to some traditional advice just ranked higher right because if you want to get that snippet 30% of the time, you’re going to be in position 3 and then there’s this diminishing curve. What I can say is that there’s still plenty of opportunities to rank lower like around 60% of the time, the snippet is sourced from position 4 to 10 and only 1% of the time is it sourced from a position lower than 10 meaning that if you want the snippet, you have to at least be on the first page and the higher you rank the better your chances of owning the snippet. So it gets back to traditional SEO.
SERP Features vs Snippets
Speaking of PAAs. So there are always other search features that we see on the SERP and we want to know how do they play along with snippets? Because given the right authoritative site and the right blend of content we can set up some really cool scenarios where you can own the snippet, maybe a video pack and you can have multiple different listings on the same SERP with a ton of vulnerable visibility. It’s a great way to sort of dominate the visibility for a certain query or a set of queries.
So this chart shows us all those different features that we see on the same SERP as snippets and the colors represent the different date ranges of data that we pulled. The chart tells a number of really interesting stories. And what we’re going to do is go through each one and zoom in to them to tell the story.
So videos back in April, 19% of the time on the same surface a snippet, we saw a standalone video pack. In November, that dove down to 2%. Just totally crashed but at the same time we saw the spike in video carousel results. So effectively nowadays, 40% of the time you see a snippet on the SERP is also going to be some form of video result whether it’s a standalone pack or it’s a carousel. This is statistically significant. It means that Google is really saying hey these two things are kind of symbiotic and video’s a great format for teaching people things. It’s a great format for instructing people like to engage with it and Google knows that and they’re pairing up a lot of snippets and videos. They’re not next to each other but they’re occupying the same space on the first page of a SERP.
So with this in mind the advice is quite simple. Consider video. Is it a way for you with your own unique strengths and your businesses and the assets that you have to also answer people’s questions instead of just written content? Because Google is showing us that they believe that there is a symbiotic relationship and there’s really good interplay between the two and customers or searchers really like to interact with it. So definitely take advantage of video if you can and it makes sense for your business.
Just remember with a carousel you’re now competing with neighbors, right? It’s not just you. So you have to take the opportunity to use those thumbnails, spice them up, make them sexy, make them more eye-catching, try to stand out from your neighbors.
Remember how I said that PAAs are on the same SERPs as snippets, right next to each other 60% of the time? This data shows us that 64% of the time, they’re just on the same SERP, even if they’re not next to each other. So basically they like to show up in the same SERP and they really like to show up together. They’re BFFs. They’re like totally interlinked and Google knows this. And then we saw this data back in April we were like, you know, the future is probably some kind of like PAA as a snippet instead of just a standalone snippet, there’s going to be a PAA where the first answer is expanded and that takes the place of the snippet itself.
It turns out Google was thinking about something like that but not quite like what we were thinking about. And we saw this. So this is a new thing that we’ve just started to see. It’s a 2-snippet SERP and the query is ‘garden needs full sun’. It’s a rather kind of oblique query. Like what is the intent of the searcher in this case? And Google is trying to hedge their bet here.
So there are two ways to interpret this. One is ‘what counts as full sun’. The other one is ‘what vegetables need full sun’. So two different interpretations and Google showing two snippets.
The advice here is pretty basic. Use those PAAs. They’re our content and keyword research gold mine. If you see a PAA on your query, go to those PAA queries and then map out the query space. You can build like a tree of queries that all descend from one root keyword. Right? And I guarantee you that if you’re interested in ranking in the snippet for a certain query, you’re also interested in ranking in the snippets for those sub queries that are listed in the PAA. You obviously have to stop at some reasonable level. You can’t just go down the rabbit hole forever, but you can build up these awesome query networks, and it’s a great way to sort of yeah, they’ll expand and map out a space that you want to be visible in.
Most boring graph on earth right? It’s actually kind of interesting.
So when we looked at snippet data over the last two years, we never saw a snippet co-occur with a local pack. It was like Google intentionally decided that for local intent queries, snippets have no place. We don’t ever want to show a snippet when someone is demonstrating local intent.
In November, they flip that up and now all of a sudden we’re seeing a lofty 1% of snippets appear on local pack queries, and that’s November.
So I saw the data and I was like is this a fluke? I asked my team to go take a look at it. They actually confirm that it’s increasing. It’s not increasing massively. But as of last week, it’s something like almost 2 or 3% I think so that’s very compelling and you can guarantee that we will be looking at this and reporting on Twitter and on our blog if we see it spiking.
Anyway, for now, if I was a local SEO, I’d be super hyped about that because I’d be looking for opportunities where I can own the local pack or be in the local pack and I can own the snippet. I would also have to have an organic ranking and that’s three listings on the same SERP. So if you’re into a local SEO, you should track and find opportunities where there are snippets and local packs because great opportunity to win a lot of eyeballs.
Featured Snippet Formatting
Okay, so one final thing in this particular section around the behavior of snippets and how they’re changing is the formatting. This is pretty familiar to you I’m sure.
We’ve all seen this, this is a paragraph snippet. It’s also my favorite snippet. It’s like the gift that keeps on giving you know Google talking about what a scraper said. It is just delicious irony.
This is a list snippet with a bonus of a shopping box above it just to prove it we don’t have our heads up our asses. We know what we’re talking about and there is indeed shopping boxes above snippets from time to time. And this is a table snippet. Someone on my team wants to buy a hovercraft and I’m super into that, we’ll find the budget.
Here’s how those formats have played out over time. So paragraphs, lists and tables, right? And we’ve seen paragraphs kind of like steady off last two runs, about the same. We’re seeing lists still increase and we think they’re taking share away from tables because we’re seeing table slide down. We saw tables spike in July and we believe that at that time we looked at the user signals and they’re like, you know, this isn’t working. It’s hard for us to get right. The content isn’t always accurate, people aren’t engaging well with it. And in fact digital assistants hate tables. I’ll get into that in a second.
So we’ve seen that slide down right? Look at this. This is terrible. Like it says view on Amazon, view on Amazon, view on Amazon, read review, product name, like it’s not a good experience. Google knows that, the searchers know that, digital assistants hate them right? We have our Google Pixel phone in our office that we use for marketing research and we just like talk at it all day and yammer at it and we spent like two days speaking queries to the Pixel phone and the digital assistant and every time a table snippet was surfaced as a response to a spoken query, it never read it out loud. And you can think that that’s actually a pretty hard problem for a digital assistant or any kind of AI to figure out like how do I read a table to somebody? Do I read it road by road, I read a column by column? It’s one of those problems that seem simple but once you dig into it, it’s actually probably really hard to figure out. So yeah tables. Not good friends with digital assistants and generally hard for people to get right and we’re seeing this decrease in them.
Featured Snippet Volatility
So that’s a whirlwind tour of how we’ve seen snippets behave over the last two years and where they’re at today.
Based on that, I want to move into a different topic. I want to talk about volatility. The idea here is that everyone’s putting all this effort and budget into winning snippets. And I want people to be able to know like if you win at how hard you have to fight to keep it and how hard is it to get it in the first place and how much shake-up is there? Because we presume that there was a lot of volatility in this space.
So here’s what we did. In April, we took our 1 million keywords. We track them in STAT. We then took all the keywords that return to snippet, right? We track them for another 19 days. So in total, we have 20 days of snippet data and some like four million snippets for us to look at.
We built these things called volatility scores and appearance scores. Volatility is actually rather simple to figure out. Volatility is, well, here I’ll show you.
There are three different kinds of volatility. There is the snippet changing URL, i.e. it changes its source, it changes its content. That’s one type of volatility.
The second type is it disappears from the SERP.
The third type is it reappears on the SERP.
Each one of these events we scored with one point. The sum of all these points is the total volatility score. Therefore the higher the score, the more volatile the snippet is.
Appearance score is rather more simple. It’s a count of the number of days out of the 20-day sample period that a snippet appeared on the SERP.
So here’s an example in action using cats, of course. So five days in a sample period. Here we can see that this snippet has a volatility score of 3 and appearance score of two. It appeared two days out of the five-day sample period. During that time, it had three changes. It disappeared, it reappeared and then it disappeared again. Makes sense?
Okay. So with that data we started plotting and analyzing and we were pretty surprised about we found because remember like I said, we thought snippets are volatile. They’re not. 68% of snippets during our time in studying them displayed zero volatility.
Now for all of you critics out there, I will say the thing that you’re probably thinking. This is a high CPC keyword set. Of course, it wasn’t very volatile. Maybe. I don’t know. You tell me, you do the research, tough guy. Like, show me.
Based on our data, we saw 68% of them showed zero volatility and that means they appeared exactly the same way for 20 days. No changes whatsoever. We thought that they’d be much more volatile.
So if you own snippets, you can kind of just relax. It means that you probably don’t have to fight like hell in order to keep them. If you don’t own snippets, you’re probably a little bit pissed to see this because you’re thinking, “Man. It’s going to be hard for me to get them.” But 32% till are volatile. That means 32% have opportunities for you and we’ll try to focus on those.
The top left corner of this graph represents healthy, like low volatility high visibility, right? So top left corner is super low volatility and super high visibility and you can see everything kind of clusters in that corner. So are snippets cluster in the low volatility section? Mostly only 28% are snippets of what we would call highly volatile and the vast majority of them like to cluster in the highly visible range. Meaning that once Google decides that they want to show a snippet, they tend not to change their mind. They’re all in on that.
The Nature of Volatility
Now snippets have these three different things that they can do to display volatility. Change the url, disappear and then reappear. And all those three things together have five potential different combinations of events or volatility types. And we wanted to know what type of volatility is responsible for most of the volatility that we see. I’m going to say the word volatility one more time, just because, volatility.
This chart shows us those different types of volatility. Right?
So we have only url changes, only disappearances where the snippet leaves the SERP and then never comes back. Disappearances and then subsequent reappearances, and then the other two types that we bucketed together and we’ll just forget about those because it’s interesting is the vast majority of volatility that we see is actually URL changes. So Google is really willing to shop around for better content i.e. better user experience, better fulfillment of the searchers needs. That’s the most common type of volatility we see.
Second interesting thing I’d point out here though is that if you win a snippet and it disappears from the SERP, it’s more than likely that it’s going to come back and when it does, it’s going to come back with the same URL that’s represented by that third bar there. I’d call it the color but I’m super color bar and I don’t want to get it wrong. So it’s like a yellowy mustard thing I think.
Anyway, okay. So if URLs are the key to volatility, we of course want to know everything we can about the urls. This chart shows something cool, which I’m just going to explain because I find these trends kind of confusing.
So what we discovered was that the vast majority of the snippets that we saw displayed up to three URLs during their time on this syrup i.e. at any given time, there are three different URLs that could potentially be in that snippet and if you own a snippet that means you’re only competing with two different sites. So the competitive field that you’re worried about narrows drastically down to just two other competitors.
76% of are snippets displayed only one URL i.e. they didn’t change but the fact that you only have to worry about two different competitors is actually like really relaxing for me like we own snippets and I think about like having to worry about everyone in the world is stealing it from us when in fact at any given time, there’s probably only two.
The advice here is really simple and is really like self-boosting for us because the best way to keep advantage of this or take advantage of this is to track daily, right? You’re only going to see those fluctuations if you’re tracking on a daily basis. When you’re able to see the snippet go to some competitor and then go back to you and then go to competitor 3 and then go back to you. So if you’re tracking daily, you’re going to be able to collect those URLs, know who you’re competing with, look at their content, figure out how its optimized, figure out how to do better than it and if you’re tracking weekly you’re going to miss all this nuance.
All right. So that’s a little bit about the mechanics there.
Spotting Volatility on the SERP
We were curious about if we could spot certain behaviors on the SERP that are either, a, tied to volatility are somehow going to indicate the volatility might happen and that’s this next section here.
One of the key indicators of volatility that we saw was format flipping. So as volatility goes from 0 to low to high, the percentage possibility that this snippet is going to return with a different format type increases. And format wasn’t part of our definition of volatility. It’s outside of that definition of volatility so we can say that if you have a snippet and it changes URLs, the chances that it’s also going to change format type from like paragraph to list or list to table or something like that increases depending on what band of volatility that snippet has. That might seem super confusing. Feel free to ask me about it in Q&A. I did my best. This is a really hard slide to talk about.
The point is that you can keep tabs on format flipping to spot volatility. So if you don’t want to go through all the effort of building a volatility score like we did which was a real pain in the butt, you can just look for changes in format because changes in format will indicate volatile SERPs and the more rapidly the format changes, the more volatile that SERPs gonna be.
The strongest indicator of volatility that we saw was this source position. The blue represents no volatility. So that’s healthy and that’s that natural curve that you guys have already seen. The orange, yeah, you know the orange and red things, those represent higher volatility and what we can say is that as source position drops down the SERP, volatility goes up.
So the advice here is super basic again, like rank higher if you want better stable snippets. So in general, rank higher. There’s some novel advice, feel free to walk away from SearchLove with it. Right?
But the thing is the higher you rank on the SERP, the more likely you are to get the snippet and the more likely you are to have a stable snippet. So it’s just healthier all over to rank higher organically.
Featured Snippets & Voice Search
Okay, so we talked about like the behavior of snippets and we know that there was more than ever we’ve talked about the stability of snippets, you know that they’re actually pretty stable. So at the end of the day we feel like this is a pretty good investment like this is worth our effort. And the third thing I want to talk about is that I’m with voice search and this is the clincher because I want to prove to you that through going after snippets, you can win the voice search race, and that’s what we’re going to talk about next.
The obvious connection is that digital assistants, most especially Google’s are reliant on featured snippets to provide easy to read out loud authoritative answers to common questions. In other words snippets impact voice search. No snippets, worst voice search experience. So create lots of snippets and win the voice search race, right? Right.
But there’s a far more interesting than you look at and that is the impact that voice search is having on featured snippets because featured snippets are changing and what they respond to and I’ll explain but first we’re just going to like go back in time.
Back when, I know it’s a long talk, I had to like make it a little spicy for you guys. So it’s not all data, there are cute dinosaur pictures too, you’re welcome.
The point is that okay, I’ll back it up.
Back in time, people used to like use keyboards to like type in queries into the search engines and when we did that, we kind of conditioned the way we searched, right? We truncated our phrases. We dropped out certain stop words. We focused on key terms. In other words, like we constructed robotic sounding searches for robotic search engines. Now overtime, search engines have evolved, so too as their ability to understand human language and importantly understand the intent behind queries. And that’s a good thing because it’s way easier for me to change the way I type and write than it is for me to change my manner of speech. And so voice search is going to be successful, search engines really have to understand normal human speech. They can’t expect people to somehow like change the way they speak right?
So what does this mean for featured snippets? It means that featured snippets have to be able to understand and respond to natural human language in order to be successful as something that’s going to support voice search. And wouldn’t you know, we’re seeing evidence of this exact strategy happening. We’re seeing more and more featured snippets appearing for longer full sentence queries than for the typical old queries that we’ve always seen.
Over the last two years, we’ve seen an increase in the number of words used in a query that surfaces featured snippets. So this is out of our data set and that means that the featured snippets are appearing for longer and longer and longer queries. When we take our terms and we bucket and average them, the terms in those long tail queries, we find the word ‘how’ appears more than any other word. Followed closely by ‘does’, ‘what’, ‘to’, ‘is’, ‘where’, these are all words that are used to compose full sentences. Full questions.
And interestingly we’re seeing how digital assistants are really really good at taking a spoken query and transposing that into a search query and I think that’s actually something that no one’s talking about enough. And so I’m going to spend some time talking about that today because there’s this awesome evidence of like machine learning being applied to this and it’s actually quite advanced. And it kind of turns keyword research on its head a little bit and I don’t know of any software tool that’s accurately helping people do this. So you kind of have to do it manually right now, unless somebody in the room wants to build it in which case I want to give you a job.
All right, so we have a Pixel phone.
By the way anyone who was at my table at dinner last night and didn’t know what the hell I was talking about after three drinks? This is the thing that I was trying to talk about last night. So, all right, cool. Here we go.
So we’re talking to a Pixel phone. I’m going to say what are the best headphones for a hundred bucks? And it makes a search query that is ‘best headphones for a hundred bucks’. And then we say, “What about wireless?” And it makes the query ‘best wireless headphones for a hundred bucks’. And then we say “I meant a hundred Canadian” and it says, it makes the query ‘best wireless headphones for a hundred Canadian’.
Not only is Google able to like accurately pop out a full sentence query out of the spoken command, but it’s able to like string together the topic across three different queries and that’s kind of amazing right? Like it through some NLP magic it knows exactly what we’re talking about. NLP, natural language processing. Anyway, it knows exactly what we’re talking about and it’s able to string together the subject from query to create a query and that’s very very impressive.
And we wouldn’t be STAT if we weren’t also hoping to like have a good laugh and focus on some of the bad stuff that we saw. So here’s a few interesting ones.
I don’t know who’s thinking about bundt cakes on my team so much but somebody’s got a problem. But anyway, how do I bake a bundt cake? Gets the query. How do I bake a bundt cake? And the follow-on query is how long will it take? And the query that’s created is how long does it take to take a bundt cake? And then we tried another one which is, “How long does that take?” And then that got accurate, “How long does a bundt cake take to bake?”
So these are the same ask, right? But in the first one Google is having a hard time interpreting what we mean by the word ‘take’. The second one they get it right. I mean the first one would be right if you were the kind of person who wanted to figure out how long it would take for you to steal somebody’s freshly baked bundt cake. You’re a terrible person if you think about that. But interesting, right?
So we’re seeing this improve quite dramatically and we’re not surprised. Google is spending tons of money on this and it’s really cool to watch it happen. It’s a bit of like a revolution in keyword research and I wish more people would focus on how spoken queries are transposed into search queries because there’s a lot of really ripe data there. It’s a really interesting space.
Here’s another example. How do I bake a bundt cake? What kind of pan does it take? How much do those cost? The third query gets transposed into ‘how much does bundt cake cost’ but that’s not what we were asking, right? So a normal human would take the word ‘those’ and would relate it back to the previous noun using the second query. Earlier, we praised Google for their ability to maintain the same subject through. But here we actually need it to be able to switch tracks and it can’t do that yet.
So while we are seeing improvements, they’re not quite there yet, but it’s happening and that’s a good thing because we did come across this one, which was “Does it take butter?” and we found ourselves looking at a SERP for ‘how do I bake butter?’ Like there’s content for this. I don’t know who’s doing that. Someone’s baking butter out there. Is their name Paula Deen? It’s probably Paula Deen.
Anyway, so what does this all mean? It means that we’re in the middle of a natural language renaissance and that pretty soon, Google’s going to have us all being like amateur linguists arguing over semantics.
On the side, the lowest point of my career is when someone on my content team had to explain this joke to me. That’s Aristotle on the left. He’s all about like allegory, and that’s Plato on the right and he’s all about like precise statements and facts and she explained it to me and I felt really chagrined and it’s very funny.
So what does this mean for voice search? Well, it means that if you are interested in snagging snippets to optimize for voice search, it means that you have to start building out lists of like full sentence queries, like full statements. You can’t, you be doing the kind of research that you used to do for typed inquiries for traditional SEO. You need to start focusing on building up these keyword lists that are like full sentences and then using those to track and optimized for because that’s where the snippets are going to be and that’s what they’re responding to.
It also means that you need to focus on intent. And that was something I talked about a lot last year. I don’t know if you saw me anywhere last year, but I spoke about search intent almost all year long the idea being that it’s better to know what somebody wants than who they are. Right? Knowledge of their intent is better than knowledge of the demographic and knowing their intent whether it’s like local or its commercial, informational, transactional is the fastest way for you to be able to create the right content at the right time answering the right questions in the right format for them and win the race. So intent is super important.
And I wasn’t arguing about, I wasn’t joking when I said we had to argue about semantics because we’re now in the space where we have to be able to tell like what do people actually mean when they query something? And off that information, we know what type of information do we have to give them and how do we have to format it in order for it to be applicable to them?
So I want you to think about researching the types of snippets that are appearing for your queries and then formatting your content to match. Meaning that Google knows a lot about this problem. Google knows a lot about the problem of matching query intent with the right content in the right format. And if you pay attention to what Google knows and what they’re seeing by tracking the SERPs you’re able to get sort of piggyback on their research. You’re able to go okay, like I know for this particular term, they often use tables and I should take that key from them and I should use tables as well. Why swim against that current?
So in our tf-idf weighted keyword list, we have a few different words that surfaced a certain type of format most often. And I’m going to go through a few of those to show you the lessons that we saw there.
So we have the words ‘does’ and ‘cost’. In our set, those surface the most paragraphs. It doesn’t take a lot of imagining to see how ‘does’ works well for paragraph answers, ditto ‘cost’.
Cost provides enough like, paragraphs provide enough space for like contacts to nuance around answers to cost questions.
For the words ‘best’ and ‘how to’, we saw the most lists and again doesn’t take a genius to figure out how ‘how to’ works well with lists. You’re giving people instructions step-by-step. Ordered or unordered lists work really well for this.
‘Best’. People want like opinions, right? And they want multiple opinions and so lists work really well for providing those. Here’s an example, ‘best kitty names’.
Ashes is a terrible name for a cat like. Hi, if you’ve named your cat Ashes, I’m sorry, you did that. It’s a bad name. All right. Please. I’m sorry.
Rates and abbreviations, right? For tables. So we saw almost tables for things like rates and abbreviations. Rates are like giving people multi-dimensional information which tables are really good at, like different rates and then different qualifiers of why the rates would be that rate depending on certain characteristics are criteria.
Abbreviations is pretty obvious. Here’s an example. This is your education for your Canadian education to pass your entrance exam into Canada. These are the abbreviations of Canadian provinces.
Wrapping It Up
Okay, that’s a ton of information. We just burned through it really quick. I’m looking at my timer and going how did I save five minutes off my practice time? But we’re going to wrap it all up here and package it up for you guys.
So what did we learn?
We learn that snippets show up a lot. Right? And we know that, we learn that they’re mostly stable. Like 68% have zero volatility whatsoever. So this tells us that they’re a safe investment for our time and our energy.
This chart shows the growth and snippet occurrence over time and I’m showing you this because I want to explain that it doesn’t mean that they’ll never go down. In February we saw this fairly significant culling of snippets for low-volume desktop queries, but the trend is on average up into the right.
Here’s an example of just how serious Google is about this. They’re actually putting two on a SERP now. So it’s not like they’re backing away from it, they’re actually accelerating even though we are seeing like ups and downs as they perfect and they call out unnecessary snippets or bad user experience. They are generally increasing their investment in snippets over time.
We also know that they’re our cornerstone for voice search. It’s actually Google’s like number one competitive advantage when it comes to digital assistance is their featured snippets. The ability for snippets to provide these authoritative easy to read out loud answers and no one else really has access to that. So that’s a huge one for Google.
So with those two things in mind, we know that this is a very safe investment and is one that we can rely on not having like Google pull the rug out from underneath us and if we go after it, we’re also going after voice search.
So this gets back to the most basic advice I can give you which is if you’re in any way investing in a snippet strategy, you need to track them. You need to be able to spot what queries snippets are appearing on, how those snippets are changing for matter, how they’re volatile, who your competition is, what type of content is winning? All of this comes with a good tracking regimen.
There are lots of great software out there that do it. Moz does it really well. STAT does it really well. There are others. So just make sure that if you’re into this, you are tracking it in some way because you’re not going to do well if you’re not tracking it.
That’s it. Thank you for your time. I hope you enjoy the rest of Rob Fest. If you, thank you.
I will say that if you want to get the content, go to getstat.com/searchlove you can download it all there. And yeah, thanks, cool.
Thank you Rob.
Woman from audience:
Woman from audience:
Okay, so few questions about formatting. First, for the video carousels. I’m curious if there, the example you showed, showed only videos hosted on YouTube. I’m curious if you’ve seen or if you have data about videos hosted otherwise.
Yeah off the top of my head and this is not authoritative and I’m sure whoever’s monitoring the STAT Twitter feed will get an answer to this through the SearchLove hashtag. But I do see other video hosting sites appear in the carousel. So it’s not just YouTube. But I would, someone’s paying attention here. I guarantee you that. I will confirm that for you on the searchlove, on the hashtag.
Woman from audience:
Okay, looking forward to it. The second question, thinking about e-commerce sites and maybe the format’s that, the formatting of the content that might capture these featured snippets, you know, I think of resource sections and blog posts, curious again if you have data about maybe which types might have better chances.
Yeah. Okay. So like the low-hanging fruit for a lot of sites are those resource sections like FAQs and things like that because naturally they’re kind of like, they’re predisposed to the content being formatted in a way that doesn’t detract from user experience. Right? Like, you know this. You can’t just chase the algorithm and make terrible experiences trying to exploit the algorithm. So you want to find content to focus on that is like a shoo-in for featured snippets in its natural form. So FAQs are like a great way to start there and we have resource sections like help sections of the site. If you can generally think about structuring your content as, based on my personas, hat tip to Dana wherever she is, like what type of questions are people going to ask and then if I construction my content in a way that really precisely answers that question. Like the question is listed and then the answer’s right there and anything you can do to make it easy for Google to confidently know that they’re pulling out the right quote, the right content to answer the right question, it’s going to go a long way for you. So, decent answer?
Woman from audience:
Great. Thank you.
Woman from audience:
Hi Rob. I think this is part of your talk was the volatility to me. They’re all so juicy. So I’m curious if you or anyone at STAT has seen seasonally present featured snippets that disappear the rest of the year? Anecdotally we have seen that with certain food products that are popular at certain holidays. Would you study that if you haven’t already?
Yes, I have we haven’t started it yet, but that sounds really interesting to me. I know that if I tell anyone on my team to do more featured snippet research I’m going to have a mutiny on my hands. But I think that like that is an area that would be super ripe because it does make sense that there would be some seasonal volatility. What would you want to know most? Like, sorry you don’t have the mic anymore. Yeah, like do you just want to know like what holidays are associated with what snippets for what queries or…
Woman from audience:
Well first of all is it true? Is it really happening that featured snippets are only showing up at certain times of the year because they’re relevant at certain times of the year. And then secondly, that would be a great add-on to learn if there are certain types that are associated with certain repeatable seasonal pack.
Yeah. It’s an interesting one. I’ll bring it back home with me and we’ll chat about it. Thanks for the tip. I like it.
Cool. There’s a couple behind you there.
Oh and this fella.
Man from audience:
Rob, thanks for being here. Local’s really important to us, which is why I’ve been a customer of yours for about a year now. Shameless plug there. Take a little coupon off my next invoice.
All right. Yeah!
Man from audience:
But I was wondering if you just had a quick example of maybe where a featured snippet of, showed up in a SERP for a local…
Like a query?
Man from audience:
Yeah. I don’t have one handy for you. But again, whoever’s monitoring our Twitter feed…
Man from audience:
If they can get that to me, that’d be awesome.
Going to make a, single you out, but we can get him an answer there. I think, right? For like an example query that triggers a local and a snippet.
Woman from audience:
No, no God. No, but if we get it, we’ll do it through the hashtags, that sound good?
Man from audience:
That’d be awesome. Thank you very much.
Man from audience:
Hey, so we’re talking about voice search like asking a simple question, you’re just getting a simple answer and you’re done. What value do you see in optimizing for those type of snippets or results?
Well, that’s a really good question. And the one that more people should be asking like just because it’s because it’s happening a lot doesn’t mean we should go after it. I think if you’re looking for like brand authority or like thought leadership in a space, that’s something that makes a lot of sense. Right? Like the more often your domain can pop up when somebody’s thinking about a problem or a product or a service or a category, the better.
And so if it happens at home seamlessly when they’re talking to their device and it happens at work when they searched it on their desktop computer and it happens during their lunch break when they’re on their phone, those are really good touch points for you building that sense of leadership and authority. And that’s one of the primary reasons I would go after it.
There’s still a whole bunch of like unresolved problems with voice search such as how the hell do you make money off of voice search? I don’t know right? So yeah, I would generally think about it as like brand exposure thought leadership plays and about like just knocking off one more touch point for people who are thinking about your product and seeing your brand in relation to that problem that they’re experiencing. Makes sense?
You got one more at the back there and then you guys can stop looking at me.
Man from audience:
Featured snippets that have a different photograph than what’s displayed with the website are becoming more and more common. Do you know why that is and also the relation between featured snippets and people also asked seems to be growing stronger and stronger.
Yeah. Okay, so I’ll answer the photograph one first. I don’t know like, I mean people are saying that that trend is actually reversing where you used to see a lot of like images and snippets from two different sources. I’ve heard anecdotally that that’s actually going the opposite direction now. We’re seeing a lot of them link up and whether that’s a result of SEO just being good at what they do or a shift in the approach. I don’t know. There has been some speculation that the thing that selects the image is actually powered by the image search algorithm and not the other algorithm. So perhaps that’s why their divergent. I just have to honestly say I don’t know and in many cases if I don’t have definitive data, I don’t want to come up and just do like speculation. I want to give like facts and I don’t have a fact for you on that one.
Why is PAAs and why RPAAs and snippets are like BFFs. Look at that query for the ambiguous query about garden needs full sun, right? How did Google know the two different ways that that query intent could be interpreted? PAAs are a really great signal for them to understand and map intent because if somebody searches for a query and then they pick another PAA result below and they spend a lot of time there and show positive signals there, they’re demonstrating that their intent was somehow divergent from the original query that they searched. So if you’re starting to see double snippet SERPs it’s because the PAAs are helping Google understand that when somebody searches for ‘garden needs full sun’, they maybe want one of two things. They want to know what vegetables need full sun or what like what constitutes full sun for a garden. And so that information I believe is sourced mainly from interactions of PAAs. That’s my best guess and again, I don’t have data on that. I don’t have a mole inside Google but that’s what I think.
Man from audience:
Well in this one day and saying we’re going to do mobile through first result or whatever result and not actually display it.
Sorry. Can you repeat the question?
There’s not a high chance of Google pulling featured snippets.
Like getting rid of them?
Man from audience:
Yeah and just doing mobile separately. I’m sorry voice search.
I believe that like the problem with voice search when it comes to search if you’ve used it on like with Siri or like the other one like Alexa, it’s totally bunk. Like it’s like, “Here’s a result I got from Bing” and it reads out like it links you to a SERP. The featured snippet approach is really solid for voice search and I don’t see that changing. So based on what I know right now, I don’t think it’s likely. But you know Google likes to make a liar out of all of us from time to time. So let’s see.
Man from audience:
Thank you so much.
Rob. Can I ask real quick about some data you’ve collected over the last couple of weeks. So we had the zero result experiment with just that, you know, Google content featured on the page. Did you see that there were SERPs that were sort of included in that set that have had featured snippets? Did you see any any drop of things?
I don’t specifically have that data. But we mostly saw those zero result SERPs for like really simple transactional task-based queries like time zone, calculator, things like that. So featured snippets aren’t really related much to the like task-oriented, single-purpose, I’m doing the search to do one quick thing, you know immediately so I wouldn’t guess so but I can look. I’ll let you know.
We can’t bait you into any wild speculation.
No, it’s like so dangerous. People like to do that. I don’t want to, you know.
That’s fantastic. Awesome. Rob Bucci, everybody.