Earlier this year, Google rolled out the Related Questions feature (AKA “People Also Ask”). If you haven’t seen them yet, related questions appear in an expandable box, mixed in with organic results. Here’s an example from a search for “Samsung Galaxy S6”:
If you click on any question, it expands into something that looks like a Featured Snippet:
Currently, Related Questions can occur in packs of between 1–4 questions and answers. Here’s an example of a box with only one question, on a search for “lederhosen”:
Once expanded, a typical answer contains a machine-generated snippet, a link to the source website, and a link to the Google search for the question.
How common are related questions?
We started tracking Related Questions in late July on the MozCast 10K, where they originally appeared on roughly 1.3% of queries. Keep in mind that the MozCast set tends toward commercial queries, and the absolute percentage may not represent the entire web. What’s interesting, though, is what happened after that. Here’s a graph of Related Questions prevalence since the end of July:
You can clearly see two spikes in the graph — one measured on October 27th, and one on December 1st. As of this writing (December 10th), Related Questions appeared on about 8.1% of the queries we track. In less than 5 months, Related Questions have increased 501%. This is a much faster adoption rate than other Knowledge Graph features.
Where do the answers come from?
When you expand a question, the answer looks a lot like another recent Knowledge Graph addition — Featured Snippets. Digging deeper, though, it appears that the connection is indirect at best. For example, here’s an expanded question on a search for “monopoly”:
If you click on that search, though, you get a SERP with the following Featured Snippet:
It’s interesting to note that both answers come from Investopedia, but Google is taking completely different text from two different URLs on the same site. With Featured Snippets, we know that the answer currently has to come from a site already ranking on page one, but with Related Questions, there’s no clear connection to organic results. These answers don’t seem tied to their respective SERPs.
Where do the questions come from?
It’s clear that both the answers in Related Questions and the snippets in Featured Snippets are machine-generated. Google is expanding the capabilities of the Knowledge Graph by extracting answers directly from the index. What may not be as clear, at first glance, is that machines are also generating the questions themselves. Look at the following example, from a search for “grammar check”:
Out of context, the question doesn’t even make sense. Expanded, you can see that it relates to a very specific grammar question posted on Quora. While the topic is relevant, no human would attach this question, as worded, to this search. Consider another example, for “cover letter examples”:
The first and last question are obviously, to a human, redundant. To a machine, though, they would look unique. To be fair, Google has come a long way in a short time — even a couple of months ago, some of these questions were riddled with grammar and spelling errors. As of this writing, I can’t find a single example of either.
Finally, there are the questions that no human would ever ask:
No rational human would ever want to know what kind of meat is in a gyro. It’s better that way.
What’s coming next?
It’s clear that Google is rapidly expanding their capability to generate questions and answers from the index. Both Featured Snippets and Related Questions have evolved considerably since their respective launches, and Google’s ability to understand natural language queries and semantic data is growing daily. It may be months before we fully understand if and how these results cannibalize organic clicks, but it seems very clear that Google no longer considers these features to be experimental and will be aggressively pushing forward question-and-answer style SERPs in the near future.