You have probably felt it: a search that used to turn up a straight answer now buries you in listicles, affiliate roundups, and pages that all say the same thing in slightly different words. Is Google search getting worse, or does it just feel that way? The question is whether that feeling holds up when you look at the actual search result quality research and user satisfaction with Google, not just the complaints in your feed.
The short answer is mixed, and that is the honest verdict, not a dodge. A 2024 longitudinal study of 7,392 product-review queries found real, measurable SEO spam problems in Google's commercial search results, and that evidence is strong. But no independent study proves that Google Search quality has declined across every query type, and satisfaction surveys point in different directions depending on who ran them and when. If you are trying to decide whether search engine quality decline evidence should change how you plan content, you need the scoped version of the claim, not the loud one.
What "Google is getting worse" actually means
People use "Google is getting worse" to cover five different claims, and they rarely separate them. It helps to name them, because the evidence supports some more than others.
The first is a result-quality claim: the pages Google ranks are less useful or more commercially manipulated than they used to be. The second is a user-satisfaction claim: people feel more frustrated using Google. The third is behavioral: searchers need more queries or more page visits to get an answer. The fourth is competitive: people are leaving Google for AI chatbots or community platforms. The fifth is a mechanism claim: some specific change, whether in ranking, advertising, or content production, caused the decline.
These claims sound like the same complaint, but they are measured completely differently. An academic study that looks at the pages Google ranked is not the same kind of evidence as a survey asking people whether they feel frustrated. Google's own documentation about its ranking systems is not independent proof of anything either. When you read a headline that says search is dying, it is worth asking which of the five claims it is actually backing up.
The strongest evidence for decline: a study most people haven't read
The best direct evidence on result quality comes from a 2024 academic paper called "Is Google Getting Worse? A Longitudinal Investigation of SEO Spam in Search Engines," written by researchers at Leipzig University, Bauhaus-Universität Weimar, and ScaDS.AI in Germany. It is the closest thing available to an independent audit of what Google actually returns, and it is worth understanding in some detail before you accept or dismiss the "Google is declining" narrative.
The researchers tracked Google, Bing, and DuckDuckGo for about a year, from August 2022 through September 2023, collecting results roughly every two weeks. That window covered eleven major Google ranking updates, including three helpful-content updates and three product-review updates, so the study caught Google mid-effort to fix exactly the problem it was measuring. The query set was 7,392 English-language product-review searches, built around the "best [product category]" pattern people actually type when comparison shopping.
For each query, the researchers pulled the top 20 organic results and checked them for affiliate links, review status, text complexity, vocabulary diversity, and other SEO-related traits. They matched links against nine major affiliate networks to detect commercial relationships, and used a review-page classifier that hit 79% accuracy.
What they found was not subtle. The majority of high-ranking product-review pages used affiliate marketing, and affiliate content showed up in search results far more than it appears on the broader web. Amazon Associates was the most common affiliate network. And there was an inverse relationship between affiliate marketing and content complexity: the more monetized a page was, the simpler and more repetitive its writing tended to be. The study also picked up a steady decline in vocabulary diversity across the retrieved pages over the measurement period, a proxy for content becoming more formulaic.
The most important finding, though, was about durability. Google's ranking updates produced real, visible improvements, but those improvements tended to fade. The researchers described it as a cat-and-mouse contest that search engines were losing over the long run: publishers adapted to each fix, and the underlying incentive to optimize for affiliate revenue rather than usefulness never went away.
That is meaningful, well-documented evidence of a real problem. It is also a narrower finding than the "Google is broken" version people repeat. The study covered product-review queries specifically, a category that is unusually exposed to affiliate incentives. It did not sample informational, navigational, local, or news queries, and it measured page characteristics rather than asking a single real user whether they were satisfied. The researchers themselves were careful to say their findings applied to commercial product search, not to Google as a whole.
So the defensible sentence looks like this: a 2024 longitudinal study found that product-review search results contained a disproportionate amount of affiliate content, and that Google's fixes for it often didn't hold. The indefensible sentence, which shows up constantly in commentary, is that research proves Google is worse everywhere. It doesn't.
What the satisfaction surveys actually say
If the academic study is the strongest quality evidence, satisfaction surveys are the messiest, because they measure feelings rather than results, and different surveys ask different questions of different people at different times.
The American Customer Satisfaction Index ran its Search and Social Media Study in 2024, based on 8,940 completed surveys collected over a year. Google's score rose a point to 81 out of 100, the highest of any search engine in the study, while the overall search category held steady at 80. Google also beat competitors on navigation, speed, content, and mobile experience. That is not a company self-reporting on itself: ACSI is an independent index with a large sample, and it points the opposite direction from the decline narrative.
A late-2024 survey from Vox Media and Two Cents Insights, covering more than 2,000 U.S. adults, tells a different story. It found that 42% of people considered search engines less useful than before, and 66% believed information quality was deteriorating. The same survey round, reported by Search Engine Land, found that 76% of respondents thought more than a quarter of Google shopping results looked sponsored or promoted, and only 14% called those results genuinely helpful.
An older 2019 survey run by Path Interactive and published by Moz, covering 1,400 respondents, actually found the opposite of what you might expect: 68% said Google's result quality had improved over time. That survey is dated now, and it came from an SEO-industry source rather than an independent research group, but it shows that this debate isn't new and that user sentiment hasn't moved in one clean direction over the years.
Put these together and you get a genuinely contradictory picture of user dissatisfaction with Google, and that contradiction is the honest finding, not a failure of the research. These instruments ask different questions, sample different populations, and were fielded at different times. ACSI measures a standardized satisfaction score. The Vox survey asks about perceived usefulness. The Moz survey asked about opinions and self-reported search behavior. None of them is measuring the same thing as the academic study, which looked at actual page characteristics rather than how anyone felt about them. Stacking these percentages into one trend line, the way a lot of commentary does, misrepresents all four sources.
Why Google's own numbers don't settle it either
Google has its own answer to this question, and it's worth taking seriously without treating it as independent proof.
Google runs a formal quality-evaluation program built around roughly 16,000 external Search Quality Raters, who assess results against published guidelines. The program distinguishes Page Quality (how well a page does what it's meant to do) from Needs Met (how well a specific result satisfies a specific search). Google says no single rating changes any page's ranking directly, and that raters represent real users' likely needs in their locale.
In March 2024, Google rolled out a core update aimed at showing less content made primarily to attract clicks and more content people actually found useful, along with new policies targeting expired-domain abuse, scaled-content abuse, and site-reputation abuse. A month later, Google said its combined efforts had produced 45% less low-quality, unoriginal content in search results than before, beating its own internal target of 40%.
That is a striking number, and it is worth knowing. It is also Google grading its own work, without the independent detail needed to reproduce the estimate or check it against user satisfaction. None of this is dishonest on Google's part, but a company's internal evaluation of its own system isn't the same evidentiary weight as an outside academic audit or a probability-sampled survey. Treat the 45% figure as what Google believes it achieved, not as proof that users experienced a 45% better search.
Why people feel search is worse even when some numbers improve
Here's where the contradictions actually start to make sense: several things can be true at once. Google can reduce a specific spam signal, and search can still feel worse to a lot of people, because feeling worse depends on more than any one metric.
More publishers are competing for the same queries than ever, which raises the odds that low-value pages end up in visible spots even when the ranking system is working as intended. Commercial incentives push product-review publishers toward content optimized for clicks and affiliate revenue rather than independent usefulness, and that incentive doesn't disappear just because one update targets it. Publishers adapt fast: a technique that gets downranked in March often returns in a modified form by June. Search tasks themselves have also gotten more complex. People increasingly search for comparisons and recommendations that take real evaluation, not a single factual lookup, so the bar for a "good" result has quietly gone up even where the results haven't changed much.
There's also a measurement gap worth naming. A ranking system can improve on the exact signal it was built to fix while a searcher's actual experience, trust, and confidence in the answer stay flat or get worse. And there's a reporting bias in every survey: people with strong complaints post about them, while people who had an ordinary, successful search don't tend to mention it. None of these explanations is proven as the single cause. They're plausible, evidence-backed reasons the "Google feels worse" experience and the "some metrics improved" finding can both be real.
What this means for how you plan content
If you run content for a team, the temptation here is to read "search quality is declining" as permission to publish more, faster, on the theory that the bar has dropped. The research says close to the opposite.
The clearest lesson from the 2024 study is that optimization and usefulness pulled apart in exactly the category researchers measured: pages got more monetized and more search-optimized while getting simpler and less distinctive. That's not a loophole to exploit, it's a warning about what not to build. The pages worth publishing now carry first-hand experience and original evidence, explain clearly how a claim was established, make a real comparison instead of restating what ten other pages already say, and are upfront about any commercial relationship. Proving firsthand experience without sounding like a marketing page is its own skill worth building deliberately.
Because most searchers still stay on the first results page (an older survey found only 7% browse further), the pages that do rank carry real responsibility to answer clearly and honestly rather than assume a reader will sort through ten similar options and pick the trustworthy one. Put your direct answer near the top, note the important caveats right after it, and separate what's fact from what's your interpretation. This connects directly to meeting the E-E-A-T bar on any topic where getting it wrong actually costs the reader something.
Automation is not the enemy here, but volume dressed up as value is. Google's scaled-content-abuse policy specifically targets pages made mainly to manipulate rankings rather than help anyone, and that applies whether or not automation was involved in producing them. Any content pipeline, automated or not, needs a real check for originality, factual support, and claims that don't outrun the evidence, which is exactly the kind of gate a human review and QA step provides. At DeepSmith we build that review step into the pipeline itself rather than bolting it on afterward, because a scaled system without a real check is the same failure mode the research just documented, just with better tooling.
Rankings alone won't tell you whether a page is actually working, since the research shows a ranking number and a real user's trust can move in opposite directions. Track whether the page answers the question without sending someone back to search again, watch engaged visits rather than raw traffic, and pay attention to qualitative feedback and complaints, not just the dashboard. This is worth building into a broader content strategy rather than treating it as an afterthought once a piece is live.
Finally, don't apply one verdict to your whole site. The strongest research signal here is specific to commercial product-review search, a category that lives or dies on affiliate incentives. A recipe blog, a documentation page, and a local-service page face different pressures entirely, so segment your thinking by query type instead of reacting to "search is declining" as a single, sitewide fact. If your content spans awareness and decision-stage pages, an E-E-A-T content audit by cluster is a more useful lens than a blanket policy change.
What would actually change this verdict
The mixed verdict would move toward "confirmed decline" if researchers ran a large, independent, multi-year study covering a representative mix of query types, not just product reviews, combining page-quality measures with direct user-satisfaction data, and replicated it across independent teams. It would move toward "not a real problem" if repeated studies found stable or improving quality across categories and the product-review findings failed to hold up on retest. Neither has happened yet. What exists right now is real, scoped evidence of a spam problem in one high-incentive category, alongside satisfaction data that contradicts itself depending on who's asking and when.
That's not a satisfying headline, but it's the accurate one, and it's more useful than a flat yes or no to "is Google search getting worse." The honest version tells you exactly where the actual risk lives (commercial, comparison-heavy search) and where the evidence simply doesn't support a strong claim either way.
