31% of customers now skip anything under 4.5 stars. The 2026 review numbers, applied to a Denver business.

The six numbers that changed

These are from the 2026 edition of BrightLocal’s survey, which has run since 2010. The sample is 1,002 US consumers. The year-over-year comparisons are BrightLocal’s own.

Two of those numbers are the headline. Nearly a third of your market now filters at 4.5 stars, and almost half filters at 20 reviews. Those are not preferences. They are the point where a customer stops reading and moves to the next listing.

Why the thresholds moved

Partly because ratings inflated. When most businesses sit between 4.3 and 4.9, a 4.0 reads as a warning rather than a good score, and consumers adjust the bar. Partly because Google made it easier to filter and sort by rating and recency inside Maps. And partly because of the AI summaries: 82% of respondents say they read the AI-generated summary of reviews that Google now shows above the individual reviews, and those summaries lean on recent, specific feedback.

Google also cleaned house. Its Maps team reported blocking or removing more than 292 million policy-violating reviews in 2025, along with 79 million bad edits and over 13 million fake profiles. Fewer fake five-stars in the pool means real reviews carry more of the weight, which is good news for an honest business that has simply not asked.

Where a typical Denver business sits

We audit Denver service businesses every week, and the pattern is consistent. A ten-year-old business with a good reputation typically has 25 to 60 Google reviews, a rating between 4.4 and 4.8, a last review two to five months old, and replies on roughly a third of them, most of which say “Thank you for your kind words!”

Against the 2026 thresholds that business is losing customers at three points. It fails the 4.5 filter for a third of searchers if it sits at 4.4. It fails the three-month recency check for three quarters of them. And it fails the reply expectation for nine in ten. None of those failures show up in rankings; they show up as a map pack position that gets fewer calls than it should.

To make it concrete, we looked at the Denver local pack for “denver seo company” on October 19, 2026. The three businesses shown had 80, 64 and 72 Google reviews, all at 5.0. Our own profile has 60 at 5.0 and was not in the pack. Reviews are not the only factor in that result, but the count gap is the one we can most directly close, which is why we practice what follows.

The plan: four targets, in order

1. Get above 4.5 and stay there. If you are below it, the fastest honest route is volume from satisfied customers, because each new five-star review pulls the average up more when the base is small. Do not try to remove legitimate negative reviews; reply to them well (next section) and outnumber them.
2. Pass 20, then aim for 50. Twenty clears the filter for the 47%. Fifty is where most Denver categories’ top-three sit, and where the AI summary has enough material to be specific about you. At one new review a week that is seven months from a standing start.
3. Keep something newer than 90 days, always. This is the one most established businesses fail. One review every two weeks satisfies the 74%; one a week satisfies the 32% who want the last fortnight. Build the ask into the job close, not into a quarterly campaign.
4. Reply to every review within a day. 89% expect a response and the same-day expectation has tripled. A reply in 24 hours is now table stakes, and it is the easiest of the four targets because it costs nothing but attention.

How to ask without being that business: at the moment the customer is happiest, in person or by text within the hour, with a direct link to the review form, one time. Google’s review policies prohibit offering incentives for reviews and gating (only asking happy customers through a filter), and its July 2026 update to review-snippet guidance added that structured data must not represent incentivized or undisclosed reviews. Ask everyone, reward no one, and the reviews you get will read as real because they are.

Replies: 50% are put off by templates

Half of consumers say a templated reply puts them off. That is the one statistic in this survey that is entirely within your control today. A good reply has three parts and takes ninety seconds:

• Name the specific thing. “Glad the new furnace kept up on Tuesday night” tells every future reader this is a real business reading real reviews.
• Add one useful fact. “The filter should be swapped every 90 days; we’ll text a reminder.” Future readers, and the AI summary, pick these up.
• Sign it with a name. Not the company. A person.

For negative reviews, the same structure plus one sentence of ownership and one of resolution, and nothing defensive. Future customers read your worst review and your reply to it before they read anything else. Our guide to responding to Google reviews has worked examples for the good, the bad and the fake.

What the AI summaries are reading

The 82% figure deserves its own section because it changes what a review is for. Google’s review summaries, and the recommendations in AI Mode, ChatGPT and Perplexity, are built from the text of reviews, not the stars. A hundred reviews that say “great service” produce a summary that says great service. Thirty reviews that mention same-day scheduling, clear pricing and a technician by name produce a summary a customer can act on.

You cannot script customer reviews, and you should not try. You can shape what they mention by what you do and what you say at the close: “If you have a minute to leave a review, it helps other homeowners to know what the job was and how the pricing worked.” Specific prompts produce specific reviews, which produce specific summaries, which produce the recommendation. Our piece on which review sites AI assistants read covers where beyond Google this matters.

The Denver competition check

Before you set targets, look at the three businesses in the map pack for your main search from your own neighborhood. Note their count, rating and most recent review date. Your targets are: match the lowest count in the pack within six months, stay within 0.1 of their rating, and have a newer review than any of them at all times. That is a plan you can run with a text-message template and a calendar reminder, and it moves the needle on the surface that produces calls.

If you want the numbers pulled for you, the free audit includes a review benchmark against your actual local competitors. If you want the system run for you, Google Business Profile management is the service.

Keep going