Rich Answers Are on The Rise!
Google’s Rich Answers: What They Are, How They Work, How to Make Them Work for You
Back in February of this year, we published a study showing how often Google responds to search queries with rich answers. So we decided to take a fresh look and see how much Google has expanded their use of these types of enhanced search results. (A “rich answer” is any attempt by Google to answer the searcher’s query in search results in a way not requiring a click through to a website.)
In today’s post, I will show you what’s changed, and speculate a bit on what this means for where search is headed. At the end of the post, I’ll also share some info on how this can benefit you.
Update: In May 2016 we ran our test of Rich Answers in search again. See how much they’ve grown and changed since this post was published.
The data for the study we published in February was pulled in December of 2014, while the data for this update was pulled in July. The total queries used was 855,243 queries, and both studies used the exact same set of queries. As a result, we have a strict apples to apples comparison. Here is the top line of the results:
Study: Google rich answers in search grew significantly over the first half of 2015.Click To Tweet
We also saw some key formatting changes. For example, rich answers that included both some text and a title went up from 2,410 to 28,884 (an increase of 778.6 percent), so clearly Google decided that including a title is very valuable.
Use of images also went up quite a bit, from 32,191 instances to 56,435 (an increase of 75.3 percent). This should be no surprise!
Back in February, I had predicted at year’s end we would be at 40 percent of our 855K queries generating a rich answer. Based on the progress through July, we still have a chance of reaching that target.
Bear in mind that the selected query set focused on questions that we thought had a strong chance of generating a rich answer. The great majority of questions are not likely to do so. As a result, when we say 31.2 percent of the queries we tested generated a rich answer, the percentage of all search queries that would do so is much lower.
Watch a Video Discussion of this Study with the Author
What Are Rich Answers?
There are many different types of rich answers. If you want to see examples of what many of these look like, you can see that in my presentation at SMX West.
That said, the types of rich answers in our study included:
- Featured snippets – these are results extracted by Google from third party websites
- Answers provided by Google – often public domain info, but sometimes from licensed data
We didn’t include videos as a type of rich answer, as these have been a part of the search results for a long time.
Knowledge Boxes vs. Rich Answers
We use the term Knowledge Box to refer to a certain type of rich answer. This is a type of rich answer like this one:
However, there are other types of rich answers that don’t come in the form of a Knowledge Box. One example is a “snippet” result. Here is an example of one of those:
In looking at the data, one of the things we tried to do was determine how common certain types of results were. So for example, in the February study, we saw 166,366 results that included a rich answer. 4,824 of those rich answer results included carousels, similar to this result:
In the current data (from July), we saw 226,941 results with rich answers, and 5,434 of those had a carousel.
In the analysis below, we will look at the rich answer results in two different ways. In our carousel example, 5,434 results in the July data represents an increase of 12.6 percent over the 4,824 carousel results we saw in the December data. We will refer to this type of measurement as the “Raw Percentage Change”.
However, we also saw far more rich answer results in the July data than the December data. As a result, 4,824 carousel results represents 2.9 percent of the 166,366 rich answers we saw in February, but the 5,434 carousel results we saw in July is only 2.4 percent of the 226,941 Rich Answer results we saw in July.
The net of this is that while we see more carousels today, they are a smaller percentage of the rich answer results mix. We will refer to this measurement as the “Relative Percentage Change.”
What Types of Answers Increased?
Here is the detailed data on the way that the rich answers landscape has changed since February:
Has Rich Answer, No Text, No Title (Simple Rich Answers)
An example of one of these types of rich answers is as follows:
Note that this does have some text in it, but it’s not in a paragraph format. This result is also interesting because Game 3 of the Stanley Cup final is not airing tonight. Yet the question asked, “Which networks aired the 2012 Stanley Cup Final?”, is in fact answered.
The growth of this type of result sports a solid 125.6 percent raw percentage change for these types of responses, as we saw 11,964 of them in December and 26,987 of them in July. However, if we look at these as a percentage of the total rich answers, the data appears as follows:
This helps us see how much Google likes these types of results compared to other types of rich answer results (the Relative Percentage Change). That total increase was 65.4 percent.
Has Rich Answer, Text, and Title
Here is an example of this type of Rich Answer
The growth here sports a whopping 1098.5 percent raw percentage increase in these types of responses, as they grew from 2,410 in December to 28,884 in July. As a percentage of total rich answers, this is what we saw:
This represents a relative percentage increase of 778.6 percent.
Directly from Google (No Attribution Provided)
Another big area of interest is how much the raw knowledge of Google is growing. These are answers that Google provides without any attribution to a third party. This is often information that is in the public domain (such as “what is the capital of Massachusetts”) or licensed by Google, such as some song lyrics (for example: “I can’t drive 55 lyrics”). The raw numbers for this also went way up:
That’s an additional 26,373 instances of these types of results. However, much of this is reflected in the total increase in rich answers. The Relative Percentage Increase in this type of result was only 14 percent.Study: The number of questions Google answers with no website attribution continues to grow.Click To Tweet
Results That Include Tabs
An example of a result that includes tabs is as follows:
These are not a particularly common type of result, but they did increase from 102 occurrences to 215 (a raw percentage increase of 111 percent and a relative percentage increase of 54.5 percent). Clearly this remains a specialized form of result used in only very specific situations, such as:
- Arby’s menu
- Shady Glen menu
- Shane’s Rib Shack menu
- Showmars menu
- Skyline Chili menu
- Smith & Wollensky menu
- Smokey Bones menu
- Snuffy’s Malt Shop menu
- Sonic Drive-In menu
Results That Include Snippets
Snippets are modified version of traditional SERP results that don’t show up as a separate rich answer box. The example we showed previously for the largest corporation in Arkansas is one such result.
In our initial study, we didn’t capture data on snippets, but in this one we did, and we found 26,349 cases of these snippets, representing 3.1 percent of all the tested queries. Some of these were accompanied by rich answers, and some weren’t:
Note that we did not track this data in the original study, so we don’t have a comparison that we can do. However, these results are particularly interesting, as they are a completely different way of extracting information from a website.
Results With Sliders
Here is an example of this type of rich answer
We saw an increase of nearly 9,000 instances of these types of results, as follows:
This represents a raw percentage increase of 133 percent in these types of results.
Results That Include Images
Increases in the inclusion of images in rich answer results were significant, as shown here:
The raw total increased by 24,244 results showing images. The raw percentage increase was 75.3 percent, but since many of these were a part of new rich answer results, the relative percentage increase was 28.5 percent.
Results With Tables
An example of a result with a table is as follows:
SERPs that included tables saw one of the largest increases in the new data:
In my opinion, this increase reflects Google’s desire to provide as complete an answer to users as possible. In the example provided above, the original question is what is Arthur Conan Doyle known for, but people who ask that question may also be interested in knowing where he was from, what he had authored, or other items, so Google chooses to provide that information in the results.
The relative percentage increase for results that included tables went up 851.7 percent.Study: Google rich answers including a table increased by a huge percentage this year.Click To Tweet
Results With Charts
Here is an example of a result with a chart, and how it looks:
It does not initially show the chart, but once you click the down arrow, you get this:
The raw percentage increase for the inclusion of a chart in the results increased more than anything else we tested (22,879 percent):
The relative percentage increase was also quite high (16,745 percent). Clearly as of the February data, the inclusion of a chart was in an experimental stage, but that has now become pretty mainstream.Study: Largest increase in Google rich answer type this year was answers with charts.Click To Tweet
Results With Forms
This is an example of a result with a form:
Forms were evidently fairly successful as well, as they saw a raw percentage increase of 161.6 percent:
The relative percentage increase was 91.8 percent.
What Types of Answers Decreased?
There were some items that didn’t increase between the two studies. Let’s take a look:
Drop in Maps (and no Rich Answer)
This is an example of a result with a map included:
This was clearly an example of a result that Google scaled back on. It saw a Raw Percentage Decrease of 30 percent.
However, in the data we are publishing today (the data we collected in July), Google did start to provide some results that included maps and a rich answer together. 475 of these were found by us, and here is an example of one of those:
There were no results that included both a rich answer and a map in the data we published in February (the December data pull).
Drop in List Ellipses
Last, but not least, we have list-style responses, many of which included ellipses in the first study data set. Here is an example of what that looks like:
These saw a significant drop in overall raw percentage decrease of 37 percent:
This represents a relative percentage decrease of 53.8 percent.
A Couple of Examples of Broken Queries
We did see a few errors along the way, such as the one about the Stanley Cup final game airing tonight that I provided above. These illustrate how hard it is to provide these types of answers. In the case of our Stanley Cup example, the question the user asked was in fact answered, but we got some extraneous additional information that was wrong. To blame here was the fact that the NBC site had not changed the title of its page after the day of the game, but it also indicates that Google’s automation is unable to recognize that the title provided incorrect or superfluous information.
We can see a different type of error when we search on a query such as, “How far is it from ___ to ___” if one of the blanks is filled with a country, such as this one:
You can see that Google has selected a point in Northern Saskatchewan as representing “Canada,” yet that is probably not the response that the user is seeking. They probably want to know the distance to the Canadian border. Also of interest, if you search on “How far is it from California to Canada,” you get the identical answer of 1468 miles, so evidently, San Francisco is used as the representation of where California is for this particular query.
Last, but not least, let’s close with a fun one:
It’s worth clicking through this one to see the rest of the answer. It’s actually extracted from a Monty Python script. That may not seem like the right answer, but chances are that it generates pretty good engagement, so Google may not be getting signals that the actual user’s question isn’t getting answered.
Can My Site Get Selected for Rich Answers?
A lot of the times when you see these rich answer results in the SERPs, you see very high authority sites like Wikipedia. That leads many to believe that the Google algo for generating rich answers is based on authority. However, we took a close look at the authority of all the domains used in the rich answers in our data set:
Not only are 54% of the domains used Moz Domain Authority (“DA”) of 60 or less, you can actually see some sites with a DA less than 20 used by Google. So low DA is not a deal killer for having your site used by Google to generate a rich answer. Note, when Google extracts a rich answer from a third party web site, they refer to this as a “featured snippet.”Is high domain authority necessary for a site to qualify for a Google rich answer box? Study gives the answer!Click To Tweet
Will Getting a Featured Snippet Help My Traffic?
Are Featured Snippets good for your site? Do you get more traffic?
To dig into this, I reviewed data from two different sites that had obtained rich answer results from Google. Here is an example of one from Confluent Forms for the query what is a RFP?. The site has a Domain Authority of 47.
David Kutcher of Confluent Forms was kind enough to share the impact on their traffic with me, which you can see here:
That’s looking pretty good. They received a nice pop in traffic to this page when they got the Featured Snippet, and it dropped back down when they lost it.
Another site shared results with us, this time for the query how to get more followers on Google Plus. They got a Feature Snippet even though their DA is 38:
How did the traffic fare to the page getting the Featured Snippet? Ben Fisher of Steady Demand was kind enough to share that data with me, so let’s take a look:
Pretty clean gain in overall traffic! Now to be fair, there are likely many scenarios where the specific pages that get shown in Featured Snippets do NOT receive increased traffic. In the case of the two examples I have shown here, from the nature of the queries, and the presentation of their result in the SERPs, it seems clear that there is more information available on the site.
The title used for the Confluent Forms result entices the user with the availability of more information (where to find them and are they still relevant). The Steady Demand result does not include the entire set of steps.
Other result scenarios may not be as positive as the Featured Snippet may provide the desired result the user wants and provide little incentive for the user to click through to the web site. I know from my conversations with people at Google that they believe that this still has a positive impact for most sites due to the branding value. This is not an impact that we have studied.It IS possible for Google Knowledge Boxes to drive more site traffic.Click To Tweet
How Can I Get a Featured Snippet For My Site?
We conducted one additional test, which was to see if we could design content to obtain a Featured Snippet for pages on the stonetemple.com web site. We started by identifying 5 common SEO questions:
- How To Implement a NoFollow?
- How To Implement a NoIndex Tag?
- How To Implement a Rel Alternate Tag?
- How To Implement a Rel Canonical?
- How To Implement a Vary:User Agent Header?
We then taped 5 videos that answered those questions, but also covered quite a bit more than just the direct answer to the question. For example, in the NoFollow video, we also answered why you would want to implement a NoFollow, when you would want to implement it, and its impact.
When we published the videos, we included a full transcript, and made sure that a simple, clear response to the main question was provided, one that would be easy for both users, and Google to find. We also shared a link to the page via Google+ and submitted it in Search Console. Three days later, two of the pages were showing a feature snippet, such as this one:
Sweet! So in short, the summary of the key steps are:
Note that the other 3 pages never did get a Featured Snippet result, but a 40% success rate is not bad!Learn how to increase the chances your content will show in a Google Knowledge Box.Click To Tweet
At the time of the initial study, I went out on a limb and predicted that Google would get to 40 percent of search results in the test returning a rich answer by the end of 2015. We are not quite to the end of the year yet, but we did see an increase from 22.6 percent to 31.2 percent.
Will we make it to 40 percent by year’s end? That might seem a tad aggressive at this point, but I do think Google will certainly get north of 35 percent by the end of 2015.
As for obtaining these for your site, my belief it’s best to accept the inevitable. Google is going to continue down this path, so you should learn to make this work to the benefit of your business.
It will bring major branding benefits, and if you design your content the right way, it can bring real and immediate traffic benefits as well.