Search-engine results pages (SERPs) traditionally had a straightforward purpose — they delivered links to users. If a user did not click any result from the page, it meant that the user didn’t see anything valuable and that the search algorithm had failed.

Today, user engagement on a search-results page cannot be so easily captured.

Good abandonment is when the user’s information need is successfully and entirely addressed by the search-results page, with no need to click on a result or reformulate the query.

Good abandonment is usually positive for the user experience: it relies on quick, easy, search-engine–generated answers to our simple information needs. (It may be less positive for websites, to the extent that users don’t visit those sites that did the work to create the information summarized by the search engine.)

There are a few cases where users’ blind acceptance of the search engine’s answers may get them into trouble. However, overall, relying on these answers and assuming a small chance of error is more efficient than painstakingly validating each of the answers.

Good abandonment forces content creators to rethink the role they play in providing information to their users.

Answer Engines

For the past 16 years, search engines have been transitioning into answer engines: we noted it back in 2004, and we’ll reiterate it now in 2020. The search engine’s primary job is no longer resource discovery; rather, it is answering questions.

Search engines like Google provide answers in various ways. (Google is by far the most popular search engine, but competitors like Bing are adopting similar approaches, as are search engines in other countries — for example, Baidu in China.) Perhaps the most extreme one is delivering the answer in search suggestions, before the user even submits a query.

A user looking to find the mayor of Kansas City didn’t have to submit a query to get her information — she didn’t even have to type the full query before the answer (Quinton Lucas) was returned in the search suggestions.
The dominant search engine in China, Baidu, has featured snippets for some queries.

But primarily, the search-engines’ transition to answer engines has been spurred on by the development of SERP features: elements on the page that served up unpaid, organic content in innovative formats, different from traditional search results. Many of these features are interactive — meaning that users can refine their query and explore alternative information without leaving the search-results page.

A search for orchids returns a SERP with many SERP features: a knowledge panel, video pack, local pack, and a People also ask element. For definitions of these and other SERP features, see The Anatomy of a Search-Results Page.

Good Abandonment Saves Time

From a user perspective, search engines becoming answer engines is generally a good thing. Human beings are satisficers: we’ll do as much work as we need to get by, but no more than that. This tendency shows up in our information-seeking behaviors. It may seem like laziness, but it’s really an efficient strategy.

Imagine you’re buying your first home, and you want to decide which type of mortgage you’ll need. You don’t need to read all the information on mortgages that is available online. (And you can’t read it all anyway!) You’ll collect information until you feel sufficiently confident in your decision.

Google caters to this behavior — it will sift through all the information in the world and deliver a set of the very best options for you to choose from. After many, many good experiences, most people have come to trust Google. We call this phenomenon Google gullibility — blind trust in Google’s results.

They trust the Google rankings so much, that they usually don’t look beyond the first page of search results. In a meta-analysis of 471 different instances of search, users only clicked through to the second page in 2% of cases.

Google takes this trust a step further: instead of delivering just the information source, it gives users the actual information, extracted from that source. When people are searching for simple facts (statistics, dates, deadlines, addresses, phone numbers, etc.), they don’t need to click on a search result anymore: Google can scan a website for that information and present it to them in a consistent, easily scannable format. If all parts work well, this process obviously saves users’ time.

Thus the terminology: good abandonment is good in the sense that it saves users time and effort — unlike our traditional conception of SERP abandonment, the user isn’t leaving because the results weren’t useful or relevant.

One participant wanted to know whether ingesting local honey could combat allergies. She read the above snippet from healthline.com. Within four sentences, her question was answered without a single click. She told us, “Yes, I always read [the snippet]. Only because it usually has the information I need without having to go elsewhere. […] If I’m looking for something quick, and really not too much background information, these boxes are very helpful.”

When Good Abandonment Goes Bad

Search engines tailor results to the query. They’re very good at returning matches to your information request. But they don’t assess truth or accuracy. This is part of the echo chamber problem of the internet: whatever most people think is good will be provided as an answer, whether it’s correct or not.

Any result that appears towards the top of a results page gets a bit of extra credibility, just because Google ranked it highly. But in many cases, had users clicked through to the result, they would have decided that the source of the information wasn’t credible. This judgment can be caused by many different (sometimes subtle) cues. Maybe the visual design of the site is poor and makes it look homemade. Maybe the content contains many unprofessional typos. Or, maybe the reader gets extra information from the page and realizes what she read on the SERP was taken out of context.

Such an assessment of information quality and credibility is often impossible when the answer is extracted and placed on the SERP. The content has been reformatted by Google, and only a short excerpt of the content is presented, even in a featured snippet.

In a recent quantitative eyetracking study, we asked participants to find the median household income in 2017, starting on a specific article on Census.gov. There was a twist to this task — the title of the article made it sound like the desired information would be there, but it wasn’t.

This scenario happens all the time on the web when the information scent is misleading — you expect an answer to be on a page, but it isn’t. 7 participants realized that the answer wasn’t on the page and gave up, failing the task. But the rest (30 participants) tried to locate the information by searching either on Census.gov or on Google. 18 participants found an answer through good abandonment on Google, by looking at search suggestions, featured snippets, or People also ask elements.

One participant looked at Google’s featured snippet to find the correct answer, $61,372.

Of those 18 that found their answer through good abandonment, 5 found the wrong answer. In all of those cases, the reason they got an incorrect number for the median household income in 2017 was because Google returned a slightly wrong answer and the participants didn’t read closely enough to catch it.

For example, one participant googled medium household income in usa 2017. Google caught his typo and corrected it to median. It returned a featured snippet from Wikipedia, which read:

$59,039

The U.S. Census Bureau reported in September 2017 that the real median household income was $59,039 in 2016, exceeding any previous year.

The text clearly means that $59,039 was the median household income for 2016 and was reported in 2017. But Google’s featured snippet misinterpreted the information and gave $59,039 as the answer. Two participants saw this featured snippet in their searches. Both of them looked only at the bold text, and then confidently gave it as their answer.

Two participants didn’t read the featured snippet closely, so they accidentally reported the median household income from 2016, not 2017.

Of course, you might say, this was a situation where the participants were looking for a quick fact and they had low motivation (they were only doing it because we asked them to). So it’s not such a big deal that they ended a session believing that the wrong information was correct. And, after all, the two numbers were in the same ballpark.

But think about the dozens of times in your day where you might be in that exact same situation — you want a specific piece of information and your motivation is low. Maybe you want to know the weather in Boston, the definition of archetypal, or the time zone of Croatia. Those are precisely the scenarios where you’re most likely to settle for a “good enough” answer from a Google SERP feature.

In most cases, those answers are probably correct, and if they’re incorrect, the stakes are low anyway. But those low-stake situations when you think you’ve gotten the right answer will only bolster trust in the search engine and might make you less likely to question it when it actually matters.

This raises an ethical question for designers: We spend so much time making things as easy as possible for users — is there a line we cross? Are we chipping away at people’s capacity for critical thinking or judgement?

Takeaways for Site Teams and Content Creators

Beyond the potential problems that good abandonment might pose to information seeking, it also doesn’t sound all that beneficial to the sites behind the links that users aren’t clicking. Don’t despair, there are steps you can take to adapt to the new reality of answer engines.

Clicks Aren’t Everything

SERP features are stiff competition. According to website-analytics company Jumpshot Inc., desktop searches end without a click to a website about 35% of the time (up 9% since 2016). That proportion rises to 62% on mobile.

That sounds scary if your team measures success by incoming traffic from organic search or clickthrough rates. However, there’s value in your users seeing your result, even if they don’t click on it. People often discover new sites, brands, products, and services via web search. In our eyetracking research, we’ve frequently seen users who look at a result on a SERP but don’t click on it. However, some return to the SERP and select that result or are primed to recognize that source and click it in future SERPs. Even if your users are just reading an excerpt of your content in a featured snippet, that action can build awareness for future interactions.

Claim Your Knowledge Panel

Depending on the business, some organizations and individuals can claim ownership of their knowledge panels. If your company hasn’t already done this, it’s an important step. When you’ve claimed your knowledge panel, you can adjust what images and facts are highlighted.

For businesses that do have a knowledge panel, the reviews pulled from Google and Facebook are often placed directly beneath important information like hours of operation. That should be yet another motivation for businesses to listen to the voice of their customer.

Compete by Answering Questions Well

SEO is a fast-paced field where the rules of the game are constantly changing. Search engines constantly update their algorithms, so know that SERP features are volatile. Moreover, because of the nature of how the featured snippets are designed, not every business can reach featured-snippet status. If every result stands out, no result stands out. As a consequence, businesses are jockeying for the few spots on the page.

The best way to compete is by providing high-quality answers to common questions related to your business or topic. Answer the Public is a great resource for uncovering common questions around a specific subject.

Answer the Public reverse-engineers Google queries to provide common questions around a topic. If your company installs solar panels, for example, it might be worth writing content that answers common questions like “how are solar panels made” and “which solar panel is best of a motorhome.”

Make sure you know your competition. Look at the sites that currently occupy a featured-snippet space that you’re after and examine how they’re formatting their content.

Keep an open mind to the different mediums in which you want to be accessible to your audience. The content from featured snippets often feeds voice search, so study what your customers are searching for with voice search and how they are saying it. Searches on voice assistants tend to prompt natural-language answers. The largest proportion of featured snippets structured as queries are prompted by how and what is questions.