Reputation Physics

The Single Star Rating Is Retiring

For years, the number a buyer checked first was the average star rating. That is no longer the whole test. Current survey evidence shows buyers now evaluating reviews across at least four separate properties at once: how recent the reviews are, how many exist, how the ratings are distributed, and how fast new ones keep showing up. Pew Research Center's benchmark 2016 study, still the most recent nationally representative read on this question, found 82% of US adults consult online ratings and reviews before a first-time purchase, and 40% do so nearly every time, so the habit itself is not new. What has changed is what buyers are checking for once they get there. A single aggregate score used to be enough of a proxy. Now it is one input among several, and the newer ones (freshness and volume in particular) appear to be hardening fast.

The number buyers check first is no longer the average. It's the date.

Buyers Want This Week's Reviews, Not Last Year's

BrightLocal's 2026 survey of 1,002 US adults found that 74% of consumers want reviews from the last three months before they trust them, and the demand for genuinely fresh reviews is accelerating. Consumers wanting reviews from the last two weeks rose from 20% in 2025 to 32% in 2026, and consumers expecting a review posted within the last week rose from 6% to 19% over the same year. That is roughly a tripling of the strictest freshness demand in a single year.

This figure comes from one commercial panel and has not yet been independently corroborated by a second source, so it should be read as a strong directional signal rather than a fixed number. But the direction is unambiguous: a review that felt current a year ago may already read as stale to a meaningful share of today's buyers. Reputation is starting to behave less like a static asset you build once and more like a stream you have to keep feeding.

Volume Became a Floor, Not a Bonus

The same 2026 survey found 47% of US consumers will not consider a business with fewer than 20 reviews, and only 9% will accept five or fewer. For a large share of buyers, review count is not a nice-to-have that improves conversion at the margin. It is a pass or fail gate that determines whether your business gets considered at all.

Older, independently established sales data backs up why volume matters so much in the first place. The Medill Spiegel Research Center at Northwestern University, studying real purchase data across two CPG retailers spanning 22 categories and more than 100,000 SKUs plus a third retailer with over 15 million annual page views, found that a product with 5 reviews saw purchase likelihood 270% higher than an identical product with zero reviews. The benefit tapers sharply after that first handful. Together the two findings suggest most of the value of review volume is front-loaded into simply clearing a credibility threshold, not in accumulating thousands more.

The Perfect Score Reads as Suspicious

Star rating expectations rose just as fast as freshness expectations. BrightLocal found 68% of US consumers now require a minimum 4.0-star rating before they will use a business, up from 55% in 2025, and 31% require 4.5 stars or higher, up from 17%. At first glance that looks like a straightforward push toward perfection. The sales data says otherwise.

Medill Spiegel's research found purchase likelihood actually peaks at star ratings of roughly 4.0 to 4.7, then declines as a rating approaches a flawless 5.0. A spotless record reads less like proof and more like a tell. Pew's research offers a plausible reason why: only 51% of review readers think reviews generally paint an accurate picture, while 48% say it is often hard to tell whether reviews are truthful. Buyers have learned to be skeptical of anything that looks too clean, which is also why Medill Spiegel found a verified-buyer badge lifts purchase odds by 15%. Buyers are not asking for perfection. They are asking for proof.

A perfect five doesn't read as proof anymore. It reads as a tell.

Negativity Carries More Weight Than Positivity, in Money and in Attention

Pew's research found 54% of Americans who read online reviews say they pay more attention to extremely negative reviews when deciding, against 43% who pay more attention to extremely positive ones. That is a stated preference, the kind survey respondents can misreport. What makes the pattern harder to dismiss is that it shows up in real transaction data too, decades earlier.

Judith Chevalier of Yale and Dina Mayzlin of USC, in research published through the NBER and the Journal of Marketing Research, tracked real book sales across Amazon.com and BarnesandNoble.com and found that a one-star increase in a book's average rating correlated with higher relative sales, and that the sales impact of one-star reviews was larger than the sales impact of five-star reviews. In other words, a bad review moves real revenue more than a good review moves it, in the opposite direction. That finding is nearly two decades old, but nothing in the newer survey or sales data available today contradicts it. If anything, the rising skepticism buyers report toward polished, uniformly positive review sets suggests the asymmetry has only become more relevant, not less.

The FTC Closed the Side Door

Every property buyers now demand (freshness, volume, and a rating that looks earned rather than manufactured) used to have a shortcut. A business could gate out negative reviews before they posted, pay for a burst of five-star reviews to look active, or stand up a fake "independent" review site. On August 14, 2024, the FTC finalized 16 CFR Part 465, a rule that closes those shortcuts directly, banning fake reviews, undisclosed insider reviews, fake company-run independent review sites, fake social media indicators, review gating, and undisclosed compensation for reviews. The rule took effect October 21, 2024, and carries civil penalties up to $51,744 per violation.

This was not a rule written in the abstract. The FTC had already built the enforcement record that justified it. In 2020 it reached a consent agreement with Sunday Riley Modern Skincare after the company's CEO directed employees to post fake five-star reviews of the company's own products on Sephora.com using accounts that concealed their identity. In 2022 it settled with Fashion Nova for $4.2 million over allegations that its third-party review system auto-posted four- and five-star reviews while withholding hundreds of thousands of lower-starred ones, the agency's first case built specifically around concealing negative reviews. Read against the current survey data, the timing is not a coincidence. The properties buyers now grade a business on are the exact properties a business used to be able to fake, and the tactics for faking them are now the specific tactics the FTC made illegal.

The properties buyers now grade you on are the exact properties the FTC just made it illegal to fake.

Then the Machine Started Reading Reviews Too

The audience checking your reviews has grown. BrightLocal found that use of AI tools such as ChatGPT to find local business recommendations jumped from 6% of US consumers in 2025 to 45% in 2026, a large enough move to make AI the third most-used discovery channel behind Google and Facebook in a single year. Whatever an AI engine says about a business now reaches nearly half of US consumers looking for a local recommendation.

Early correlational research offers a first, imperfect look at how these engines seem to be reading reputation. Insites studied 10,000 US businesses across ChatGPT and Perplexity and found that businesses visible to both engines averaged 133 Google reviews, against just 11 for businesses invisible to both, a 12x gap. Confirmed presence in Google's Local Pack correlated with 3.5x more AI recommendations. Technical SEO factors including Core Web Vitals performance, structured data, llms.txt, and FAQ schema showed weak or no correlation with AI visibility at all. SOCi's separate research found these engines are far more selective than traditional search to begin with: ChatGPT recommended only 1.2% of business locations studied and Perplexity 7.4%, against 35.9% visibility in Google's local 3-pack, roughly a 30x selectivity gap.

Read together, the two studies suggest the answer an AI engine gives about your business may lean more on raw review volume and a confirmed local footprint than on the freshness, distribution, or velocity properties human buyers are increasingly demanding, and considerably less on the technical polish of your website than most businesses assume. But both studies are correlational, run by a single vendor each, and neither one, nor any independent academic or government source we could locate, actually measures how these systems weight recency or the pace at which new reviews arrive. That is not a small gap. It is the honest state of the evidence: this layer is a frontier that requires primary measurement of your own market, not a settled formula anyone can hand you yet.

What This Adds Up To

Put the pieces together and a business's reputation is no longer one number on one platform. It is a live, multi-property record, checked on at least two different kinds of surfaces (traditional review platforms that human buyers read directly, and an AI-answer layer that appears to be reading a narrower, more volume-and-presence-weighted version of the same signal) and it is being checked under standards that are visibly tightening year over year, not standing still.

The practical implication is that a business cannot manage this by checking its star rating once a quarter. It needs a current read of where it stands on freshness, volume, distribution, and confirmed local presence, on the surfaces that actually matter for its own market, updated often enough to catch the kind of one-year swing this study documents in the recency and AI-adoption data alone. That is the map we build and call the Visibility Corpus: not a single score, but a picture of where a market's attention actually sits right now, and where it is moving next.