The Case Against Customer Reviews: The 4.7 Ceiling

A perfect five-star rating makes you *less* likely to win the sale. That is not a contrarian pose — it is what Northwestern University's Medill Spiegel Research Center found when it analysed real transaction and review data across thousands of products. Purchase likelihood climbs steeply as ratings rise, peaks somewhere in the 4.0 to 4.7 band, and then falls again as the average approaches 5.0. Shoppers read perfection as a warning sign.

Customer reviews work, but they plateau. Northwestern University's Spiegel Research Center found purchase likelihood peaks between roughly 4.0 and 4.7 stars and falls as ratings approach 5.0, with most of the lift arriving by the fifth review. JustMarket.Me argues your next marketing hour belongs to the website those reviews send people to.

Table of Contents

The short version

• Customer reviews are a qualifying asset, not a closing asset. They get you shortlisted; something else wins the job.
• The curve flattens fast: most of the lift lands in the first five reviews, and purchase likelihood declines past 4.7 stars.
• Reviews you host on your own site cannot produce stars in Google. That has been documented policy since September 2019.
• Since October 2024 in the US and April 2025 in the UK, aggressive review generation carries real fines.
• The uncapped asset is the page the review sends people to. The Trust Handoff model below shows which stage is starving.

Do customer reviews increase sales?

Yes — sharply at first, then barely at all. The Spiegel Research Center's analysis, published in How Online Reviews Influence Sales, reports that purchase likelihood for a product with five reviews is 270% higher than for the same product with none. The marginal value of review six, and every review after it, is a fraction of that.

Two more findings from the same work sharpen the point. Reviews matter more for expensive, considered purchases than cheap ones: Spiegel reports a 190% conversion lift for lower-priced items against 380% for higher-priced ones. And simply labelling a review as coming from a verified buyer improved purchase odds by 15% — a formatting change, not a review-count change.

Read those three numbers together and a shape emerges. Reviews are a threshold you cross, not a slope you climb. The business collecting its fifth review is buying something enormous. The business collecting its two hundred and eleventh is buying almost nothing, and is paying full price for it in staff time, software fees and follow-up emails.

Where does the review curve actually flatten?

At roughly five reviews for volume, and just under 4.7 stars for rating. Past those two points you are optimising a variable that has stopped moving.

The 4.7 finding is the one people refuse to believe. Spiegel's researchers attribute it to a credibility effect: consumers treat a flawless average as evidence of filtering rather than evidence of quality. In no product category the team examined was the optimal rating a perfect 5.0.

This has an uncomfortable implication for the standard review playbook. Chasing the highest possible average — asking happy customers only, timing requests around good outcomes, quietly discouraging the disappointed — moves you toward the exact number that costs you conversions, and toward practices that are now illegal in two major markets.

What changed in 2024 and 2025?

Three things: a United States federal rule that took effect in October 2024, a United Kingdom ban that took effect in April 2025, and platform enforcement that removed over 240 million reviews from Google Maps in a single year. Together they reprice the whole activity.

The United States Federal Trade Commission's Rule on the Use of Consumer Reviews and Testimonials (16 CFR Part 465) took effect on 21 October 2024. It bans fake and AI-generated reviews, reviews from company insiders that are not disclosed, and — the clause that catches ordinary small businesses — incentives conditioned on the review expressing a particular sentiment. Civil penalties run up to $51,744 per violation.

The United Kingdom went further on 6 April 2025. Under the Digital Markets, Competition and Consumers Act, posting or paying for fake reviews became a banned practice outright, and the Competition and Markets Authority's fake reviews guidance (CMA208) extends to how reviews are handled: hiding negative reviews, or presenting a star rating that gives an inaccurate overall picture, is covered. Fines reach 10% of global turnover or £300,000, whichever is higher. In practice, "we only ask the customers we know are happy" is now a compliance question, not a tactic.

Meanwhile the platforms industrialised their own enforcement. Google reported removing or blocking over 240 million policy-violating reviews on Maps in 2024, alongside 12 million fake business profiles, with most of the reviews taken down before anyone saw them.

And the quiet one, which predates all of it: Google's review snippet documentation states that when the entity being reviewed controls the reviews about itself, pages using LocalBusiness or Organization structured data are not eligible for the star rich result. The testimonial wall on your homepage, and the review widget embedded beside it, cannot produce stars in Google's results — and have not been able to since September 2019. A remarkable number of businesses still pay monthly for that widget.

What is the Trust Handoff?

The Trust Handoff is a three-stage model of how a small business turns reputation into paid work. Trust is earned somewhere you do not control, handed off through a channel you do not control, and spent on a page you control completely. The three stages are Earn, Hand off and Catch, and only the last one has no ceiling.

Earn is where nearly all small-business marketing money goes, and it is the stage with the lowest ceiling and the newest legal risk. Hand off is the stage nobody owns, and it is getting narrower as AI answers absorb clicks. Catch is the only stage with no ceiling, no regulator and no gatekeeper — and it is the stage most businesses have not touched since the site was built.

The model explains a pattern that otherwise looks like bad luck: the business with the best reviews in town that still loses jobs to a competitor with worse ones. Its trust is being earned and then dropped.

What does a dropped handoff look like in practice?

It looks like two dental practices with the same map-pack position and very different websites. Practice A has 4.9 stars from 212 reviews. Practice B has 4.6 stars from 38. On the Earn stage, A has won decisively, and A's owner has the invoices from a review-generation subscription to prove it.

Now the handoff. Both appear in the same map pack for the same search. Say 100 people tap through to each site on a phone. A's site is a five-year-old template: it takes 4.8 seconds to become usable on a mid-range handset, has no online booking, shows the phone number inside an image, and keeps opening hours in a PDF. B's site becomes usable in 1.4 seconds, puts "Book a check-up" above the fold, and shows the next three available slots.

Nobody can tell you the exact conversion gap between those two sites; anyone quoting a precise figure is guessing. What *is* measured is the sensitivity. Google and Deloitte's Milliseconds Make Millions study of 37 brand sites and more than 30 million user sessions found a 0.1-second mobile speed improvement associated with an 8.4% lift in retail conversions and 10.1% in travel. Those are retail and travel figures, not dental ones, and they describe one tenth of one second. Practice A is behind by more than thirty of those, before you count the missing booking button.

Practice A's next review costs money and moves a number that stopped mattering at review five. Its page speed and booking flow cost money once and move the only number without a ceiling. This is why the return on a website rebuild is easier to argue than another year of review software, and why the websites we build for dental practices start from the booking flow and the served HTML, not a testimonial carousel.

What is the strongest counter-argument for reviews?

The strongest counter-argument for customer reviews is that they are the only marketing asset a small business owns which cannot be bought outright, compounds indefinitely, and keeps working while the owner sleeps. It is a genuinely strong case, and it deserves stating properly.

Google's own local ranking guidance says prominence is based partly on how many reviews you have, and that "more reviews and positive ratings can help your business's local ranking." Reviews are therefore not only a conversion asset but a visibility asset — they feed the stage before the handoff. AI answer engines lean heavily on aggregated third-party sentiment when they summarise "best plumber near me", because that sentiment is the only thing in their index that is not written by the business itself. And a business with four reviews is filtered out by human beings before a website is ever loaded, no matter how good that website is.

All of that is true. None of it contradicts the argument here. The claim is not that reviews are worthless — it is that review *maximisation* is the wrong plan past a threshold that arrives far earlier than the industry selling review software would like you to know. Cross the threshold: get to a healthy volume, keep the rating honest, reply to the bad ones in public. Then stop, and spend the next hour somewhere the ceiling has not already been hit.

One honest exception: if you have fewer than roughly ten reviews, or a genuinely poor rating, none of this applies to you yet. Fix that first. Everything above is written for the business that already won the Earn stage and cannot work out why the phone is quiet — the same failure mode described in the case against landing pages.

So what should you actually do?

Run the 60-second handoff test, then act on what it tells you.

1. Take a phone off Wi-Fi. Search for your business the way a stranger would — service plus town, not your business name.
2. Tap through from the listing to your site and start a stopwatch. Time how long until you can *read* something useful, not until the spinner appears.
3. Try to book, quote or call in three taps. Count the taps. If you cannot do it in three, neither can a customer holding a screaming toddler.
4. Ask what proof is on that first screen. Prices or honest ranges, service area, credentials, opening hours as text. A review carousel is not proof if the page cannot answer what you do and where.
5. View the page as text only. If your key content is missing, Googlebot and every AI crawler are seeing what you are seeing — the single most expensive fault a small site can have, and the one covered in the local SEO playbook for dentists.

If steps two to five go badly, you have found where your reviews are going to die, and it is not on the review platform. Fixing it is a one-off cost with a permanent effect, which is the opposite of the review treadmill. JustMarket.Me publishes what a site build and ongoing optimisation cost on its website and SEO pricing page, and the same logic applies whether you build with us or with anyone else: the Catch stage is the only one you own.

The uncomfortable summary: your reviews are a promise made by other people, and your website is where you either keep it or break it. Most businesses are spending everything on the promise and nothing on keeping it.