Review trust collapsed from 79% to 42%. Your responses just became the product.

In 2020, BrightLocal asked consumers whether they trusted online reviews as much as personal recommendations, and 79% said yes. In the 2025 edition of the same survey, that number was 42%. Not a dip - a collapse. Half the trust, gone in five years, in the single most-cited statistic the local marketing industry has ever produced.

Most coverage of that number treats it as bad news for everybody. It isn’t. It’s bad news for star ratings and very good news for one specific group: businesses that write real responses. Here’s the thesis - as review trust falls, response trust rises, because responses are the one part of a review profile that can’t be bought in batches of twenty.

What actually broke

Three things happened to review credibility at once, and readers noticed all of them.

Fake reviews industrialized. Google removed 240 million policy-violating reviews in 2024 and 292 million in 2025 - against roughly a billion submitted. Sellers openly price negative-review batches at $100 per twenty, and the 2025 extortion wave (review-bombing followed by a WhatsApp ransom demand) made national news. Consumers didn’t read the transparency reports, but they saw the pile-ons and the too-perfect profiles, and recalibrated.

AI made volume free. The marginal cost of a plausible five-star review dropped to zero, and everyone knows it. Google explicitly banned AI-generated review content and built classifiers to catch it, but the reader’s heuristic updated faster than any classifier: generic praise now reads as synthetic by default. “Great service, highly recommend!” ×40 used to look like a beloved business. Now it looks like a purchase order.

Incentives burned the well. A decade of “leave us a review for 10% off” - now an FTC violation with penalties north of $50,000 per instance under the Consumer Reviews Rule - taught consumers that positive reviews are often bought, gated, or nudged. Even honest five-stars inherited the discount.

Where the trust went

Readers didn’t stop reading - BrightLocal’s own data shows 71% still read reviews regularly, and usage keeps growing. They changed what they weight. Ask people how they evaluate a business now and the answers cluster around signals that resist manufacturing:

• Recency and cadence - a steady drip of reviews over years beats any total. Cadence is expensive to fake and cheap to earn.
• Specific complaints, handled visibly - the 4.5 with a few detailed problems reads as more trustworthy than the 5.0, because real businesses have bad Tuesdays.
• The responses. ReviewTrackers’ research puts it plainly: 97% of people who read reviews also read the business’s responses. Ninety-seven percent. The response column is not an afterthought - it’s co-equal content with the reviews themselves, and it’s the only column you author.

Why responses resist the rot

Think about what a skeptical reader can and can’t discount. A five-star review might be bought. A one-star might be a competitor - bought batches and competitor fakes are both documented industries now. But a response thread where the owner names the specific failure (“we double-booked our only mobile tech”), states the fix, and signs their name? There is no marketplace for that. You can’t buy twenty of them for $100, because each one has to engage with a real complaint’s specifics, in a consistent voice, over months. Consistency across time is precisely the thing fabrication is bad at.

This is the same reason the fakes are detectable in aggregate: fabricated content regresses to the generic. It’s also why the worst thing you can do in the low-trust era is respond with templates pasted verbatim - a wall of “we’re sorry to hear about your experience” reads exactly like the synthetic reviews readers already discount. (Templates as scaffolding are fine; that’s what a tone-matched starting point is for. Templates as the finished product are self-sabotage.)

The strategic consequence: reallocate the effort

Most small businesses allocate reputation effort roughly like this: chase the rating, worry about the bad review, dash off thank-yous. The 42% number says that allocation is backwards now. Concretely:

Stop optimizing the decimal. The rating difference between 4.6 and 4.8 persuades almost nobody anymore - skeptical readers treat both as within the margin of manipulation. The visible difference is in the response column, where a specific, signed, operationally-literate reply is unmistakably human.

Treat every response as portfolio content. The 97% figure means your responses get read at nearly the rate your reviews do. Write the important ones - recent negatives, your most-visible old threads - the way you’d write a page of your website, because functionally that’s what they are.

Let the imperfections stand. Suppressing, gating, or lawyering away every negative review was always against platform policy and is now federally risky - and in the low-trust era it’s also bad marketing. Visible flaws with visible handling are the trust signal. The flawless profile is the suspicious one.

Keep the cadence machine running. Asking happy customers for reviews - plainly, at the moment of finished work, with no incentive attached - is still legal, still effective (83% of people asked actually write one, per BrightLocal’s 2026 survey), and still the only sustainable way to keep the recency signal alive.

The part nobody tells you

The trust collapse is regressive: it punishes businesses that coasted on accumulated stars and rewards businesses willing to do visible, ongoing work. That’s rare good news for small operators, because visible ongoing work is the one arena where a six-bay auto shop can outcompete a national chain. The chain has more reviews. It will never have an owner who writes, “that was our scheduling mistake, here’s what changed, ask for me when you come back.” In a market where readers have learned to distrust everything that scales, the thing that doesn’t scale is the moat.

Reviews became cheap, so readers stopped pricing them. Responses stayed expensive - a few honest minutes each - and expensive signals are the ones that still trade at par. Spend accordingly.