Is Google’s engagement-hacking masking stalled growth?

Mike Blumenthal and I have highlighted Google’s efforts to keep local searchers on Google, or within the Google ecosystem, in numerous fora for the last decade. In the early days, this mindset seemed primarily based around Google maintaining its share of eyeballs amidst the mobile insurgency of Facebook, and apps more broadly.

As it relates to local search, in these earlier days–an era of mobile-unfriendly or nonexistent business websites–constraining a searcher’s engagement within Google Business Profiles provided a more “efficient journey” (term used below) for local searchers.¹

But increasingly, as many of the DOJ discovery documents in the recently-decided antitrust case have highlighted (including the now-famous “code yellow” email from Ads boss Jerry Dischler), keeping users on Google isn’t solely about retaining market share, and it’s no longer about providing a more efficient journey: it’s about increasing the odds they will perform additional searches. 

Not to put too fine a point on it, but in the eyes of Google executives, additional searches mean additional bites at the ad revenue apple. Even if ad click rates per search go down as a result of more impressions, ad click rates per session could actually be rising as query refinements drive more precisely-targeted ads. And any raw increase in ad clicks of course sends more money to Google’s coffers.

As AJ Kohn points out in last year’s brilliant piece It’s Goog Enough, Google has achieved “full market saturation.” There might not be any internet-connected person on earth who hasn’t at least tried Google. Datos’ recent Consumer Search Behavior report confirms this premise: the percentage of users who searched using Google was largely unchanged from May 2023 to May 2024.

Having maxed out on user growth, Google’s executives have backed the company into a corner by seeming to anoint query volume as the product team’s north star metric–emails from the DOJ discovery documents suggest it is at the very least the de facto predictor of the company’s revenue. That in turn has led to some bizarre and downright suboptimal user experience decisions.

Journey Optimization vs. Engagement Hacking

Query volume is at best orthogonal to user experience, and is more frequently opposed to it. A perfect search engine can infer exactly what a user is looking for, even from an imperfect query, and return the best result first for that query. 

By canonizing query volume, not result quality, a “successful” search session for Google is one that requires multiple query refinements–hardly an optimized session from the user’s perspective.

And optimizing for query volume may partly explain (along with risk to revenue) Google’s aversion to experimentation with radically different SERP experiences of the kind we’ve seen from ChatGPT and even the kind initially suggested by Gemini.

As Ed Zitron reports from those DOJ discovery documents, at least as far back as 2019, the search team led by Shashi Thakur and Ben Gomes

wasn’t able to finely optimize engagement on Google without “hacking engagement,” a term that means effectively tricking users into spending more time on a site, and that doing so would lead them to “abandon work on efficient journeys.”

Google’s PR team highlighted former Gomes’ testimony in a response to longtime industry observer Barry Schwartz that “We have no way of growing queries directly unless we do a better job with search.”

While that may have been the search team’s stance while Gomes was in charge, it’s clear that many inside Google, presumably including SVP Prabakhar Raghavan, felt differently. At some point in the interceding years between 2019 and 2024, the engagement hackers won the internal battle vs. the journey optimizers.

Blatant Engagement Hacking

In one of his 2019 emails, Gomes actually foretells a core component of Google’s 2024 strategy (emphasis mine):

We could increase queries quite easily in the short term in user negative ways (turn off spell correction, turn off ranking improvements, place refinements all over the page)...

Refinements are now indeed all over the page. Each click on a refinement now generates a brand new query, a brand new URL, and a brand new mix of both organic and ad results.

This is true for ecommerce queries:

When I do the same faceted search on Amazon, my query remains the same. 

This is true for local business queries:

When I do the same faceted search on Yelp, my query stays the same.

This is true for travel queries:

Strangely, Google’s own “Things to Do” module shows Knowledge Panel results inline–I’d predict it’s only a matter of time before those generate new queries, too.

People Also Search

Google is so desperate for additional queries that they’re now devoting precious above-the-fold real estate not just to People Also Ask, but People Also Search, even for competitive commercial keywords.

More Subtle Engagement Hacking: “Just Goog Enough” as a Feature, not a Bug

Again from Gomes’ 2019 email (emphasis mine):

We could increase queries quite easily in the short term in user negative ways (turn off spell correction, turn off ranking improvements, place refinements all over the page)...

In fact, Eli Schwartz highlights that Google did exactly that: intentionally making its rankings worse as an experiment in 2020, with no impact on revenue.

While I don’t believe Google is intentionally making its results worse four years later, query volume as a north star creates the perverse incentive to water down results’ quality.

Results that are too obviously awful might turn more and more searchers running to the open arms of the competition–particularly early in the ChatGPT hype cycle. 

But if the experience is just bad enough, a searcher becomes the proverbial frog being boiled, gradually becoming aware that Google results aren’t as good as they used to be. 

Instead of completely changing their habit, though–a very hard thing to do–they simply perform more and more query refinements in the Google search box, exactly as the product team was hoping.

In fact, per Sparktoro, unsatisfied queries make up nearly 40% of “zero-click searches,” where the user turns right back around to the top of their browser and conducts another search.

One of the big beneficiaries of that behavior has been Reddit, but instead of just going to non-search alternatives like Reddit.com, a significant number of Google users have simply have simply begun appending “reddit” to more and more of their queries–in many cases either before or after they try the same query without the “reddit” brand appendage.

Advanced Engagement Hacking: Ad Space Re-Sale

No SERP real estate is spared in Google’s all-out drive for more queries – even SERP real estate paid for by advertisers.

For users who click on an ad and then return to the SERP, Google has long suggested People also search for boxes beneath the preceding sponsored result. This visual barrier has the effect of discouraging users from viewing organic results (and cheaper advertisers) for their original query in favor of beginning a new query, where more ad slots can potentially be sold.

But in one of our most recent search behavioral studies, we saw a jaw-dropping “enhancement” popping up for a noticeable subset of the 20% of users who clicked on ads in a competitive local category.

Our research methodology looks at the position of a given click (among dozens of other attributes), and we were confused to see users recording multiple clicks on the same ad position without changing their query.

Upon closer inspection, we realized that if a user hits the back button in their browser, Google sometimes rotates different advertisers’ ads into the same position as the user’s original click. Ad rotation doesn’t happen every time, and it may be part of an ongoing test, but a video showing this behavior is below. Note the domain and creative switch in ad position 3, and the addition of a fourth ad. 

Perhaps Google’s rationale is that a user who clicks back to the SERP must be dissatisfied with the advertiser result. This rationale is seriously flawed, as in high-consideration categories, users will frequently evaluate multiple businesses prior to choosing which one or ones to contact.

A handful of users in our study seemed unable to find their original first choice, as Google had sold the space out from under the original advertiser to the next-highest bidder. But hey, at least they collected a few extra euros from the search session.

From I’m Feeling Lucky to I’m Feeling Like I Need to Refine My Query

Far from “organizing the world’s information,” and “doing no evil,” 20 years after its IPO, Google’s motivating principle seems to be “predictable growth for Wall Street.”

As a result, it provides an experience just bad enough that users only occasionally find what they’re looking for on their first query, but just good enough that users don’t switch to another search engine. 

Google undoubtedly has tons and tons of smart engineers and product people, and I don’t lay the blame for this strategic direction at the feet of any rank-and-file Googlers. In fact, Datos’ aforementioned study shows just how effective Google’s product team has been at achieving the goal set for them by their superiors: the average number of searches per desktop user increased by 10.4% between May 2023 and May 2024.

But strategically, what a weird goal to set. It’s led to

• substandard user experiences, 
• inhibition of true innovation, and 
• an increasing reliance on self-preferencing.

The dream of a perfectly-efficient “I’m Feeling Lucky” search journey has been set aside in favor of keeping a flat or shrinking userbase seeing as many Google screens as possible per session. It’s an all-out effort to maximize the sheer mathematical odds that any given user, on any given query, might, at some point, even accidentally, click on an ad.

Analysts predicting a decline in Google’s market share have been made to look foolish in every previous instance over the past decade, and I’m not going to do it here. Habits are very, very hard things to break. But two conditions of this particular moment feel qualitatively different: Judge Mehta’s looming antitrust remedy, and the rise of viable search alternatives with radically different (i.e. spartan) interfaces.

In this particular moment, I’m not sure Google can afford to focus so much of its product energy on gaming its numbers–a focus that is already showing mixed returns on Wall Street based on the reaction of investors to recent earnings calls. It needs a major product reset that hews to the original promise of I’m Feeling Lucky, delivering the best (third-party) result instantly, not bloated, misinformed AI-organized pages engineered to induce as many query refinements as possible.

1 This constraint is far less defensible today, given the prevalence of bugs, hacks, and review fraud rampant within Google Business Profiles and quantum leaps forward in the quality of content management systems most widely used by small businesses.