Zero-click searches, Google replies to the study

Reading time : 7 minutes

Less than 24 hours: this was the (little) time Google needed to reply to the results of Rand Fishkin’s study on the zero-click trend in Search in 2020, of which we have given news in recent days. And as we already noted in our piece, the focal point on which Danny Sullivan invites to reflecton behalf of the American group concerns the actual validity of the data analyzed by Sparktoro and Similarweb, which “are based on an imperfect methodology that misunderstands the way people use research”.

Google replies to the study on clickless searches

Thus writes the Public Liaison for Search in the post published on The Keyword, which wants to “provide an important context on misleading claims” present in the research – that, Moreover, it has also continued to generate negative criticism from the international SEO community.

And Sullivan starts right from the surveys highlighted by “research professionals” to the thesis of Fishkin – summarized in Google steals clicks to sites because “most searches on Google ends without anyone clicking on another site”, a trend that has been called zero-click searches – to challenge and resize it because it results from an imperfect methodology.

In fact, the Googler replies, “Google Search sends billions of clicks to websites every day and since Google was first created we have sent more traffic every year to the open Web“, and also “we connect people with companies in a wide variety of ways through Research, for example by allowing a direct phone call”.

How people use Search today

Sullivan’s goal is to contradict SparkToro’s arguments with facts that clarify the situation and provide a relevant background to the misleading allegations of the study, analyzing the actual context of today’s searches and therefore the different ways compared to the past in which people use Google, which can lead to search queries that require an immediate response and do not require a click.

Today, in fact, “people use Search to find a wide range of information and Google Search sends someone to a website billions of times a day”; so although not all queries return a click on a site “there are many very good reasons why this happens“, as exemplified in the following four cases (not adequately considered by Fishkin, implies even too covertly Sullivan, in certifying that “only 35% of research in 2020 led to a click”).

  1. People reformulate their questions

When they start the search, people do not always know how to formulate their questions: they could “start with a broad search, such as sneakers and, after examining the results, realize that they actually wanted to find black sneakers“.

In this case, says Google, such searches “would be considered as zero-click because they do not immediately produce a click on a site”. In the case of the purchase of sneakers, some zero-click searches may be required to get to the goal, but “If someone eventually ends up on a retailer’s site and makes a purchase, Google has brought that site a qualified visitor, less likely to come back unsatisfied”.

According to Sullivan, this situation happens frequently and, therefore, Google offers many features (such as “related searches”) “to help people formulate their searches and get the most useful result, which is often found on a site”.

  1. People search for quick facts

People look for “quick and concrete information, such as weather forecast, sports results, currency conversions, time in different places and more”. Today Google – “like many search engines”, the article notes – provides this information “directly on the results page, drawing on the licensing agreements or tools we have developed”. These results are “useful for users and are part of our ongoing work to improve Google Search every day”.

In 2020, for example, “we showed factual information on important topics such as COVID and the US elections, which generated some of the most important interests we have ever seen on research”. The feature on election results has been viewed billions of times and provided “high-quality information in real time as people waited for the result,” and Google has also “provided concrete information on COVID symptoms in collaboration with WHO and local health authorities, making critical information readily accessible and supporting our responsibility in the fight against potential online disinformation”.

  1. People directly get in touch with a company 

Compared to local businesses, Google offers “consumers many ways to get in touch directly with businesses via Google Search”, which often do not require a traditional click. For example, “people might look for opening hours, then go to the store after having confirmed that it is open, or find restaurants on Google and call for information or to place an order, using the phone numbers listed by us”.

On average, Sullivan reveals, local search results generate “more than 4 billion connections for companies every month”, including over 2 billion visits to sites and connections such as phone calls, directions, food orders and reservations.

Moreover, “we help the many local businesses that do not have their own website”: through Google My Business “companies can create and manage their own page on Google and be found online”, and every month Google Search “connects people with over 120 million companies that don’t have a website”.

  1. People directly access the apps 

Some searches lead people directly to apps, rather than sites: for example, “if you are looking for a TV program, you will see links to various streaming providers such as Netflix or Hulu”. So, if the user has that streaming app on the smartphone they can directly open the link in the app, and the same goes for many other apps like Instagram, Amazon, Spotify and others.

More opportunities for sites and companies

Sullivan continues to describe the positive aspects of Google in connecting users to websites, products and activities, recalling that “every day we send billions of visits to websites and the traffic we sent to the open Web has increased every year since Google Search was created”.

In these 20 years and more “we have worked to constantly improve Google Search by designing and implementing useful features to help people quickly find what they are looking for, including maps, videos, links to products and services that you can buy directly, flights and hotel options and local business information such as opening times and delivery services”.

In this way, Google has “greatly increased the opportunities for sites to reach people” and now the SERP, “which previously showed 10 blue links, shows an average of 26 links to sites on a single page of mobile search results”.

Google is more of a simple link list

In conclusion, says Sullivan, although “some argue that we should return to showing only 10 blue links to websites” Google has taken a path to improving the information shown to users: although “we show links to sites for many queries when they are the most useful answer“, at the same time “we also want to create new features that organize information in more useful ways than a simple list of links”.

The continuous introduction of these features over the past two decades has also led to the growth of “traffic that we are targeting on the Web, demonstrating that this is useful for both consumers and businesses”.

The critical points of Google’s defense

It is clear the Google’s goal (at the hands of Danny Sullivan), which from the beginning aims to debunk the claims of Sparktoro and, in particular, to challenge the data on the progressive decrease of traffic to sites due to the so-called zero-click trend, which would be “stealing website visits“.

Yet, even Danny Sullivan’s article is not exempt from providing partial or contextless information, as noted by Barry Schwartz and Roger Montti (who are also skeptical of Fishkin’s research, however).

In particular, some points of the defence are criticized:

  • Google claims to have increased the number of visitors lead to sites each year, but in this way substantially blends numbers proportional to absolute numbers indistinctly.
  • Google states that reformulation and refinement of queries “happen often”, but does not say how often.
  • Google explains that people use search to find “factual information”, but it does not reveal the percentage.

Moreover, regardless of a growing overall search volume, it seems clear that Google is actually working to capture and own an increasing share of traffic, implementing features and solutions that make sure that a high percentage of users never leave Google or a property owned by Google. And nothing in Sullivan’s piece disproves or contrasts this feeling.

Therefore, the data present in the article “are not enough”, Schwartz rightly says, which specifically asks Google to open “tracking of featured snippets in Google Search Console” because, not showing such data, “Google seems to be hiding something from publishers”.

While not “fully in agreement with the Sparktoro study” because he believes that “the data in that study should be broken down and analyzed with greater detail and precision”, Barry Schwartz is convinced that even Google “should share more data with publishers and communicate how Google Search is benefiting the publisher, not just users who search”.

The most relevant issues of the SparkToro study

From this story no winner emerges and the SEO community is actually skeptical about both positions expressed by Google and SparkToro.

Said of the limits of Google’s defense, are particularly strong accusations made to the study of the company of Rand Fishkin, which would be driven by a cognitive bias, or an error of assessment or lack of objectivity of basic judgment.

In particular, Sparktoro’s analysis would be the victim of availability bias, “a distortion that results from the use of more readily available information, rather than necessarily more representative ones”. In this case, the data taken into consideration for the study may not be a true random sample (despite the large number of queries examined) because they only come from the “Similarweb’s proprietary panel of tens of millions of users who have installed their apps”.

This is a problem already highlighted for the previous Sparktoro survey of 2019 – then based on Jumpshot data that, according to some observers, used a sample not really random and not sufficiently representative of all searches on Google, as it comes only from “users using the Avast antivirus”. This was then characterized as “a classic sampling error, usually referred to as availability bias”, a situation repeated even in the present case.

In conclusion, we always need a context

What we can say, as a partial conclusion of this controversy, is that data and statistics without context should always be taken with a grain of salt.

In the specific case of Google searches, we should analyze what kind of questions people ask and how they have changed over time, and then specifically study what are the types of searches that lead to zero clicks and check if there is a legitimate reason for the absence of a subsequent click to a non-google site.

Only by understanding what are the current search contexts and deepening the evolution of user behavior can you really have a picture of the situation; for example, searches for a phone number of a company or the text of a song (which are search contexts) are changing and are increasing because more people rely more on mobile devices, and therefore “the claim that Google is stealing clicks does not hold“.

However, as search marketers the controversy interests us marginally and what matters is to focus on getting the best and most useful traffic for our sites and customers, using Sparktoro’s study information and Google’s response to understand a little more how the Search works and which direction it is taking (without thinking of a “right” or a “wrong”).


Call to action