If you follow me on LinkedIn or you've read this blog in the last weeks you know that I've been experimenting a lot with the featured snippet. In fact, as part of the team at WordLift in the previous year, I've been training dozens of people on how to integrate structured data and open linked data in their SEO strategy.
As of 2018 structured data and open linked data are two crucial factors that Google is taking into account to rank web pages. Not only that. In fact, nowadays things are changing, and Google's search algorithm is getting smarter and smarter. For that matter, there are three crucial aspects of search to take into account:
- Google's featured snippet
- Google's Knowledge panel
- Voice search
From a superficial look, those might seem like three separated aspects of SEO. However, that is not the case. In fact, with a proper Semantic SEO strategy, you can target them all.
A little caveat: the purpose of this article is to show you the findings of my experiments. I try to test things, not in the sense of gathering data but rather to do simple experiments with a low probability of succeeding. When one of those does succeed; for me the signal is clear. There is something in the experiment that Google liked. I can't say exactly why and what. But that doesn't matter. As long as you follow the steps I've taken chances are you'll also get similar results.
Start with the end in mind
A few weeks back on this blog I published an article about Google's business model:
I intended to target a long-tail keyword with low search volume and competition and see if I could trigger a featured snippet. Once and if the featured snippet would have been triggered. I'd wait a few days and then test whether that also triggered a voice search from my Google assistant. That is what I did, and it worked!
In short, the question I targeted was:
"what's a hidden revenue business model?"
After 24 hours from publishing the article associated with that query, the featured snippet appeared:
After a few days that featured snippet has become a trigger for the voice search answer you get from Google's assistants:
How did that happen?
Dissecting Google's featured snippet
Going from the featured snippet to voice search is not automatic. Backlinko recently analyzed over 10,000 Google Home search results, and it found that 40.7% of all voice search answers came from a Featured Snippet.
Of course, thinking about a fixed percentage might be deceiving as Google continuously updates its search algorithm. Also, voice search is so new that it's hard to say how it will evolve in the next months.
Yet as of now a featured snippet is a powerful way to get into voice search, that is also why I like a strategy based on targeting for the Google's featured snippet.
In a recent post I explained how I triggered the featured snippet and what steps I did take to make that happen:
Above you can see a summary and breakdown of five critical aspects. I suggest you read the article below to have a more in-depth overview:
Google's search algorithm is smart enough to split text and images: how to optimize for both image and text in the featured snippet
In my editorial strategy, you'll see that in some cases I divide up the title of an article in two parts. The first part is usually a question. In fact, the purpose of that question is to target a featured snippet.
Recently I had written an article on DuckDuckGo business model:
I was targeting the question: "How does DuckDuckGo make money?"
However, this one was way more competitive as the featured snippet was already taken by Quora (as you might imagine Google trusts Quora way more than a small blog like mine). However, not all is lost.
Thus, I had already an answer to that question on Quora. So I used the infographic I had produced for my article, and I added it to the same question on Quora.
My reasoning was simple. First, the question that was ranking first on Quora and that Google was using as featured snippet didn't have any image inside. So why not try to see whether I could trigger a featured snippet by signaling Google that my answer was more comprehensive as it comprised the infographic.
Second, by positioning my content on Quora, I was also trying to signal to Google that the same content coming from my blog could be trusted. In short, I was trying to use Quora as a vehicle to bring credibility toward my blog. Something interesting happened.
The day after I had repurposed my content over Quora. When I typed on Google the question "how does DuckDuckGo make money?" it triggered the old featured snippet.
This time though, while the text was coming from the old answer; the image was coming from my answer (which includes a graphics I had inserted inside). Even though I didn't manage to steal the featured snippet; I still managed to position my image inside the featured snippet.
In short, I believe since the text answer from Quora didn't provide any graphics. The search algorithms pulled it out from mine.
In other words, Google's search algorithm is smart enough to split the featured snippet in half: text and image.
Remember the purpose of the featured snippet is to give a direct, short answer to users for questions. Further, the goal is to provide relevant content that can quickly answer, be it text or an image.
That is also why at times the search algorithm might get the text from a site and the image from another site. Therefore, when you're planning to trigger a snippet, you have to make sure to do a consistent strategy for both text and image. Both have to be prone to be "snipped!"
Do you want to skip the line? Target the featured snippet!
From the data above you can see that also search results that are at the 4th or 5th position have chances of getting the featured snippet. This is important because if you do get it, then you'll steal traffic from the first position.
As pointed out by the same study, when a search result is ranked first, and there is no featured snippet, it gets 26% of the traffic for the overall query. When instead, there is a featured snippet on that page the first result only reaches 19.6% of traffic. In other words, the featured snippet steals traffic from the first positions.
At times a featured snippet might be triggered by the last results on the first page, just like it happened to me with this query:
Even though my result was in the 7th position in the SERP (therefore according to ahrefs.com study it only had 0.7% of chances of getting snipped) it still triggered a featured snippet.
Featured snippet vs. knowledge panel: Who's the winner?
Thus, control my brand through Google. Therefore, I set up a page for the scope and used a strategy that I explained in this article. This SEO experiment is important because usually a featured snippet of a person is triggered either when there is a Wikipedia page that supports it. Or the site from where it's coming from has high authority. Yet I wanted to see whether I could do that with my small blog based on the quality of data I provided to Google.
It did trigger it:
Thus, the featured snippet showed up. Also, something interesting happened! A knowledge panel that before wasn't there was triggered:
In this particular case that is pulling up information about me taken from an author page, I have on Amazon.
What is a Google Knowledge Panel?
When people search for a business on Google, they may see information about that business in a box that appears to the right of their search results. The information in the box, called the Knowledge Panel, can help customers discover and contact your business.
How do you get one?
Like search results, whether or not a business’s information will appear in the Knowledge Panel is determined by a variety of factors. Relevance, distance, and the prominence of the business all contribute to its standing in local search results. Verifying a business does not guarantee that it will appear in the Knowledge Panel.
In short, it seemed like the work I've done for the featured snippet (Schema and Open Linked Data might have been the most significant contributors) had also triggered the knowledge panel.
However, as of the time of this writing, the knowledge panel has eaten the featured snippet:
Like in a poker game where there are winning hands. If I had to compare featured snippet to knowledge panel; I'd say the featured snippet is like having a four of a kind. Instead, the knowledge panel is like having a straight flush!
In fact, today if you look for "Gennaro Cuofano" you'll only see that box on the right side of the search box (that is the knowledge panel).
As we've seen the knowledge panel has a different purpose than the featured snippet; while the featured snippet aim is to answer a specific question. The knowledge panel gives relevant information about a business.
The featured snippet is way more fluid and based on rankings; the knowledge is based on the ability of the information you've provided to enter Google's knowledge vault. While there is no sure way to trigger a knowledge panel; getting a featured snippet might facilitate it. When the knowledge panel cannibalizes the featured snippet that is a good sign - I argue - as this implies that Google might be trusting more your data.
As pointed out by Dr. Peter J. Meyers on Moz, over time featured snippets get cannibalized by knowledge panels:
It's likely that Google is trying to standardize answers for common terms, and perhaps they were seeing quality or consistency issues in Featured Snippets. In some cases, like "HDMI cables", Featured Snippets were often coming from top e-commerce sites, which are trying to sell products. These aren't always a good fit for unbiased definitions. Its also likely that Google would like to beef up the Knowledge Graph and rely less, where possible, on outside sites for answers.
Therefore, over time Google's knowledge panels might consolidate around websites like Wikipedia, Amazon and so on.
What's going on with Google?
A debate is going on about what's happening with Google. In fact, Google introduced a featured snippet with no SERP:
In other words, for some queries, Google prefers to give a result in the form of a featured snippet. If users want to see the other search results they have to voluntarily click on "show all results." This seems to be an important change as for the first time Google is hiding its results page.
On the one hand, this isn't surprising as those queries (like knowing what's the time) have a clear intent. In short, users just want to get a straight answer. In this case, Google is acting just like Wolfram Alpha. In addition, as pointed out by Bill Slawski this isn't new as it was since 2005 that Google started to work on "Just the Facts, Fast":
— Bill Slawski ⚓ (@bill_slawski) March 15, 2018
Yet now the question comes naturally: Will Google cannibalize content from the SERP? Is going getting into the content business?
Well, I thought what a better way to solve the impasse than to ask Google itself. If you think about it - I argue - it's all about its business model. If the business model fails, then the company doesn't exist anymore.
So I asked Google itself, what is its business model:
The answer is clear: "Googe is a one-stop shop for helping you find things."
Now, "one-stop shop" means that users can get all they need right there, on that blank page. Thinking about how Google has evolved in the last years. There is no doubt that Google needed third-party websites - it still needs them - to offer relevant content to its users, besides paid ads.
Imagine a web filtered through Google where all you got was a bunch of paid ads. Who would have found that interesting at all? Instead, when Google introduced its paid network for publishers (AdSense), this allowed any site to monetize their content quickly.
As anyone - that deals with Google - knows, the search engine from Mountain View loves content. This spurred the birth of a content industry on the web, which main aim was to feed as much web pages to Google. The deal - although implicit - seemed to work quite well. Google got "quality content" from third websites. It allowed them to monetize it; while those websites remained the intermediaries between Google's users and Google itself.
This deal made sense for anyone. In fact, Google didn't have enough power to produce content itself. However, starting 2012 Google has introduced a set of initiatives (Knowledge Graphs, Hummingbird, and RankBrain just to mention a few) that made it way smarter than it was. Those initiatives allowed Google to gather content around the web, converts it into data, which can be easily manipulated to create new data. That data, in turn, can become content, served to its users.
Thus, if Google can produce that content itself? Would it still make sense to show search results coming from third-party websites?
The question remains open. It's too early and too hard to say what will the future hold. There are a few considerations to make though. First, if we look at AdSense compared to AdWords. There is no doubt that the former help fuel the latter. However, it is true that Google shares most of its revenues with AdSense partners. That is also why AdSense has such lower margins compared to AdWords.
Thinking business, if I were Google, I'd try to focus on the part of the business that has high margins. Second, Google has no control over third-party websites part of AdSense. As a company that wants to be a one-stop shop, it also makes sense to have as much control over its content. Thirds, today for Google might be cheaper to index the visible web and give back search results. But what if it becomes cheaper to create its own content? Imagine how much resources would Google save in terms of crawling budgets, and spamming controls.
Another little caveat: featured snippets are volatile as they depend on ranking and other aspects that Google might take into account from time to time. Thus, some snippets that appeared at the time of this writing might have disappeared. However, the overall strategy proved to be successful, at least in the short run. In fact, the objective isn't just the featured snippet but the knowledge panel, which is usually way more stable over time. Yet, one and the other might often be connected if in you use a Semantic SEO strategy based on structured data and open linked data.
Numbers and Results
Let's talk numbers now!
As you can see even though the number of total impressions deteriorated by 18% (from 31,170 impressions down to 25,525). On the other hand, if we look at total clicks, they went up by 6% (from 521 to 554) in the last 28 days. In addition, the average CTR (click through rate) has increased from 1.52% to 2.17% (this is a staggering 42% increase). Also, the average position of my keywords on the SERP increased from 47.8 up to 45.9.
If we look at the numbers in absolute terms, they might seem small. Yet, if we take into account the % increase that is not bad at all! Imagine this kind of metrics improvement for a large site.
If you look at the users for this month (in blue) from organic traffic alone, compared to the previous month (in orange) that is clear that the slope is upward. I also noticed that some of that traffic was coming from specific questions:
On those keywords the volume is low. Yet as of now they seem to have a good click-through rate. Of course, numbers are still too small to draw any conclusion. However, It will be fun to see how traffic coming from those keywords will evolve over time.
Now it's time to go back and make more SEO experiments!