Reviews are a distribution channel
Seventy-four percent of consumers want reviews from the last three months. That makes reviews a flow you have to run, not a total you accumulate.
Reviews are usually treated as a score you accumulate and then possess. The data says the opposite: three quarters of consumers want reviews written in the last three months, which makes your rating a perishable flow rather than a stored total. A business with 4.9 stars and forty reviews, all from 2022, is not well reviewed. It is invisible.
The numbers, and the one that changes the model
BrightLocal surveys around a thousand US consumers on this each year, and the February 2026 edition is worth reading properly rather than in the summarised version that circulates.
The headline items are the ones everyone quotes. Ninety-seven percent read reviews for local businesses. Forty-one percent say they always read them, up sharply from 29 percent the year before. Sixty-eight percent will not use a business rated below four stars, up from 55 percent, and 31 percent now insist on 4.5 or better, nearly double the 17 percent of a year earlier. Forty-seven percent will not use a business with fewer than twenty reviews.
Set beside the previous year, though, those figures say something the summaries miss. The bar is not high. It is rising, fast, and in one year:
| What consumers say | A year earlier | Now |
|---|---|---|
| Always read reviews before choosing | 29% | 41% |
| Will not use a business under 4 stars | 55% | 68% |
| Insist on 4.5 or better | 17% | 31% |
| Always write one when asked | 16% | 28% |
Every row moved in the same direction and none of them moved slightly. The share demanding 4.5 or better nearly doubled in twelve months. A business that was comfortably above the threshold last year can be below it this year without its rating having changed at all, because the threshold came up to meet it.
The fourth row is the one to sit with, because it is the only one that is good news and it moved just as far. Customers are not merely more demanding, they are also considerably more willing to write. The same shift that raises the bar also hands you the means to clear it.
That recency finding is what turns this from a reputation topic into an operating one. A rating is not a possession. It decays, in the sense that a body of excellent old reviews stops doing the job somewhere around the three month mark, and the business that was well regarded in 2023 has to earn the same standing again this quarter.
Which reframes the whole activity. If reviews were a total, the correct approach would be a push: a campaign, once, to get the count up. Because they are a flow, the correct approach is a rate: a small number arriving every week, forever, which is a process rather than a project.
Why almost nobody has one
The gap here is not effort or reluctance. The same survey found that 83 percent of consumers who were asked to leave a review went on to do it, and 28 percent now say they always write one when asked, up from 16 percent.
Eighty-three percent. There is no persuasion problem, no incentive required, and no clever wording that unlocks it. The overwhelming majority of customers will do this on request. The entire deficit is that in most service businesses nobody asks, because asking is not anybody's job and there is no moment in the process where it happens.
A channel where 83 percent of prompted people convert, running at almost zero volume, because the prompt does not exist anywhere in the workflow.
The second reason is timing. The person best placed to ask is the technician who just finished the job, standing in front of a satisfied customer, and that person is already thinking about the next appointment. Asking feels awkward, it is not on the job sheet, and nothing in the system prompts it. So the request either happens through individual initiative or not at all, which in practice means not at all.
Treating it as a channel, in the operating sense
If this is a channel, it gets the same treatment as any other: a target rate, an owner, a trigger, and a measurement. None of that requires software beyond what the business already has.
Set a rate, not a total
A number of new reviews per week, sized so the most recent ten are always from the current quarter. For most small service businesses that is somewhere between two and five a week, which is a fraction of completed jobs rather than all of them.
Know what the rate actually buys, because it is less than it sounds
One a week puts thirteen inside the ninety-day window and two inside the last fortnight, which clears both recency bars, and reaches the twenty-review threshold in twenty weeks. Two a week halves that to ten. Five a week gets there in a month. The floor is roughly one, and everything above it is margin and speed rather than necessity.
Attach the ask to an event, not a person
The trigger is job completion or invoice payment, automatically, not a technician remembering. Anything that depends on somebody feeling like it will run for three weeks and then stop.
Ask in the channel the customer already uses
A message to the phone that placed the booking, with a direct link. Every additional step, a login, a search, a form, removes a share of the 83 percent, and the losses are entirely mechanical.
Respond to all of them, within a week
Eighty-nine percent of consumers expect a response, 80 percent say they are more likely to use a business that answers all reviews, and 81 percent expect the answer inside a week. This is cheap and almost universally skipped.
Read the negatives as operating data
A complaint about a missed appointment window is not a reputation problem, it is a dispatch problem that has published itself. The review is the symptom and the schedule is the disease.
Responding to every review has the best ratio of effort to effect of anything on that list. It costs a few minutes a week, it is visible to every future reader, and the expectation is now near universal. A business with an unanswered wall of reviews is telling every prospective customer something about how it handles communication, and the message is not the one it intends.
What this is worth, and what it is not
Being precise about the mechanism matters, because reviews get oversold as a growth strategy and they are not one.
What they actually do is convert demand that already exists. Somebody searching for a service in your area will see several options and will use reviews to eliminate most of them. Strong recent reviews mean you survive the shortlist. They do not create the search, they do not make anybody need a new roller shutter, and they will not fill a diary that is empty because nobody knows the company exists.
That said, the multiplier is large and the cost is near zero, which is an unusual combination. There is no media spend, no agency, and no new capability. The customers are ones you already served and the work is already done. That places it in the same category as the unanswered phone: money already earned and not collected, which is why both belong ahead of anything that costs real money.
Two things not to do
Both are common and both are worse than doing nothing.
The first is the bulk push. Asking two hundred past customers in one week produces a cluster of reviews all dated the same fortnight, which reads as manufactured to a human and is exactly the pattern platforms look for. It also does not solve the recency problem, because in four months the cluster is old and you are back where you started. A steady trickle beats a spike on every dimension that matters.
The second is filtering. Asking only the customers you expect to be positive, or routing unhappy ones to a private form instead, is against the terms of every major platform and it destroys the thing that makes reviews useful. A perfect five-star record with no critical reviews at all reads as false, and the survey behaviour supports that: people go looking for the three-star reviews specifically, because that is where the balanced assessment lives.
Ask everybody, take what comes, respond to all of it. A four-point-seven with a handful of mixed reviews that were answered thoughtfully converts better than a suspiciously perfect five, and it is the only version that survives contact with a platform's enforcement. Where this sits relative to the other things worth fixing is in the four-layer growth engine, and why an owner has not got to it yet, despite it being nearly free, is the operating layer problem in miniature. The whole diagnostic is on the system page.
The short version
- The bar is rising, not merely high. In one year, those insisting on 4.5 or better went 17 to 31 percent, and those refusing anything under 4 stars went 55 to 68. A rating can fall below the line without changing.
- The same shift also made people more willing to write: 16 to 28 percent now always leave one when asked. The thing raising the bar hands you the means to clear it.
- The required rate is lower than it feels. One a week puts 13 inside the 90-day window and 2 inside the last fortnight, and reaches 20 reviews in 20 weeks.
- Seventy-four percent of consumers want reviews from the last three months and 32 percent from the last two weeks. Reviews are a flow, not a total.
- Eighty-three percent of people who are asked leave one. The deficit is entirely that nobody asks, because asking is nobody's job.
- Set a weekly rate rather than a target count, trigger the ask off job completion rather than off somebody remembering, and send it to the phone that booked.
- Respond to all of them within a week. Eighty-nine percent expect it and 80 percent say it makes them more likely to use you.
- Never bulk-push and never filter. A cluster of same-week reviews reads as manufactured, and a flawless record reads as false.
Questions I get on this
How recent do online reviews need to be?
How do you get customers to leave reviews?
Should you respond to every review?
All consumer figures from BrightLocal's Local Consumer Review Survey 2026, published 11 February 2026, surveying 1,002 US adult consumers.