Seven years, three reviews, one afternoon

A note before we start: we sell PrestaShop modules for a living. That gives us a conflict of interest in any conversation about reviews, and it is exactly why this article names no competitor, no platform as a villain, and no specific case as fake. The goal here is to hand you a method you can use on anyone, including us, not to score points against a rival.

A module had been on sale for seven years. It had three reviews. All three were posted on the same afternoon.

That is the whole story, and it is enough to make you pause. Not because three five-star reviews are suspicious, they are lovely, but because seven years of real customers almost never behave like that. Real feedback arrives the way rain does: unevenly, sometimes in a drizzle, sometimes not at all for months, occasionally in a burst after a good week. It does not wait seven years and then fall in a single afternoon.

Here is the uncomfortable turn, though: none of that proves anything. Reviews can lie, but timestamps are worse at it. They give away a pattern the words are careful to hide. A cluster of dates can have a perfectly innocent explanation. Move a shop from one platform to another and every imported review can be re-stamped with the migration date. Suddenly seven years of genuine history looks like one busy afternoon, and the seller did nothing wrong, though, as we'll argue later, clearly labelling those re-dated imports is on them.

So this is not an article about catching liars. It is an article about provenance, about learning to ask, calmly and without accusation, "where did this come from, and does the shape of it make sense?" That question is a skill. Let's build it, and then let's turn it on our own store in public.

Why this is a moral question before it's a technical one

Social proof is borrowed trust. When a stranger writes "this worked for my store," and you let that tip your decision, you are trusting someone you will never meet on the strength of someone else you will never meet. That is a remarkable thing for commerce to run on, and it only works because of an unspoken agreement: the proof is real.

Manufacturing that proof isn't a growth hack with an asterisk. It is a quiet breach of the one agreement that makes the whole system function. It spends the goodwill of the exact customer who leaned on it, the person who bought because "everyone" seemed happy, and then found out "everyone" was a spreadsheet. You can dress it up as marketing, but underneath it is the same move as any other confidence trick: get someone to rely on something that isn't there.

It also corrodes the thing you actually want. The point of a review section isn't a number; it's a signal you can act on. Fake it and you blind yourself first. You lose the honest feedback that would have told you the checkout was confusing or the docs were thin. A store that manufactures its own applause never hears the note that's out of tune.

And the weight of that trust falls hardest on sellers who control their own review stack. This is worth being precise about, because it's the heart of the matter.

Where reviews live: three trust models

Not all review systems carry the same burden, and the difference isn't about how strict any particular company is, it's structural. It comes down to one question: how much distance is there between the person who wants five stars and the button that publishes them?

None of these is dishonest by nature. Plenty of the most trustworthy review sections on the internet are fully self-hosted, and plenty of open platforms are full of genuine, hard-won feedback. The point is only this: the more control you have over the proof, the more of your own integrity is riding on it. A self-hosted widget where the same person wants the stars and owns the database carries the heaviest burden there is. Not because such sellers are suspect, but because nothing except their own honesty stands between a real review section and a fabricated one.

That this is an industry-wide concern rather than any one company's failing is easy to show with neutral numbers. Open platforms publish their enforcement figures: one well-known review platform reported removing several million fake reviews in a single year, and a major search engine reported taking down hundreds of millions of policy-violating reviews across its listings in 2025. Those are not scandals about specific villains; they're the weather. Fake social proof is common enough that the people who host reviews now remove them at industrial scale, and, as we'll see, common enough that lawmakers stepped in.

Red flags that deserve a second look

None of the signals below is proof. Each one is a question. A place where a reasonable person would want an explanation before trusting the number. Real, honest review sections trip some of these too, and that's fine. The skill is not "spot the fake." The skill is "notice what deserves a second look, and go find out."

Notice the structure of that table. Every row has a right-hand column, because every red flag has a boring, innocent twin. That column is not there to let anyone off the hook. It is there to keep you honest. The moment you find yourself certain a review section is fake, you have stopped reading provenance and started writing fiction of your own.

Same product age. Same five-star average. Completely different shape over time, and the right-hand pattern is a question to ask, not a verdict to hand down.

Read the section, not the score

The star average is the least interesting number on the page. Everything useful is in the shape underneath it, who wrote it, when, whether they actually bought the thing, and whether the seller was brave enough to leave a lukewarm review up and answer it. (If you're on the receiving end of a bad run, here's how to respond to negative reviews and recover.)

The example alongside is illustrative and anonymized. Two reviews share wording and landed the same afternoon with no purchase on file; the third is specific, verified against a real order, four stars rather than five, and the seller replied. Which one would you actually trust? The honest review is the one that admits setup took an hour. Perfection is not persuasive. Specifics are.

The number lies; the shape doesn't

If you take one habit from this article, make it this: stop reading the average and start reading the distribution. A 4.7 can be built two completely different ways. One is a real spread, mostly fives, a healthy chunk of fours, a few threes, and the occasional one-star from someone whose theme fought the install. The other is a near-perfect spike of fives with essentially nothing else, across hundreds of reviews. Same headline number. Very different truth.

Real customers are gloriously inconsistent. Some love your product and still knock a star off because shipping was slow. Some misunderstand a feature and blame you for it. That noise is not a flaw in your review section. It is the signal. A distribution with no fours and no complaints, sustained over a large sample, is the genuinely unusual thing. Not proof of anything. Just the shape that most deserves a "how did that happen?"

Both average about 4.7. The lumpy one, with real fours and a few complaints, is the one that reads as human.

Audit your own store in five minutes

The best way to learn this method is to run it on yourself, where you already know the truth. If your reviews live in your own database, a self-hosted widget, a PrestaShop reviews module, the native product-comments table, three small queries will show you the exact shapes we've been describing. Adapt the table and column names to whatever your review source actually uses; the logic is the same everywhere.

-- Review provenance self-audit
-- Adapt "reviews", "date_add", "rating", "verified_purchase", "ip_address"
-- to your own table and column names.

-- 1) Arrival velocity: reviews per month, and how many were verified purchases
SELECT DATE_FORMAT(date_add, '%Y-%m') AS month,
       COUNT(*)                       AS reviews,
       SUM(verified_purchase)         AS verified,
       ROUND(AVG(rating), 2)          AS avg_rating
FROM   reviews
GROUP  BY month
ORDER  BY month;

-- 2) Clustering: any single day with an unusual burst, from how few IPs?
SELECT DATE(date_add)             AS day,
       COUNT(*)                   AS reviews,
       COUNT(DISTINCT ip_address) AS distinct_ips
FROM   reviews
GROUP  BY day
HAVING COUNT(*) >= 10           -- tune the threshold to your volume
ORDER  BY reviews DESC;

-- 3) Distribution: is anything other than 5 stars ever present?
SELECT rating,
       COUNT(*)                                                   AS n,
       ROUND(100 * COUNT(*) / (SELECT COUNT(*) FROM reviews), 1)  AS pct
FROM   reviews
GROUP  BY rating
ORDER  BY rating DESC;

Query one tells you whether reviews arrived like weather or like a switch being flipped. Query two surfaces the "seven years, one afternoon" pattern, and the distinct_ips column is the tell within the tell: thirty reviews from two addresses is a very different story from thirty reviews from thirty. Query three shows you whether your distribution is human-lumpy or suspiciously clean.

The same three checks, run against an example store's review table (illustrative). The flagged day isn't a conviction. It's a prompt to go find the provenance.

Run this before you ever point it at someone else. If your own numbers make you wince, that's not a reason to hide them. It's a to-do list. (Ours gave us a to-do list too; more on that at the end.) For the practical side of gathering and displaying reviews well, our companion piece on where customer reviews should live on your store goes deeper, and built-in vs. third-party review systems covers how to host them in the first place.

The law caught up in 2024 and 2025

For most of e-commerce's life, faking social proof was a matter of conscience. That changed recently, and quickly. Across the three markets most PrestaShop merchants sell into, manufacturing reviews went from "dishonest" to outright illegal, and in the US, faking social-media engagement did too, with real penalties attached.

• United States, the FTC rule. The Federal Trade Commission's Rule on the Use of Consumer Reviews and Testimonials came into force on 21 October 2024. It bans fake and AI-generated reviews, buying or selling reviews, reviews from insiders that don't disclose the connection, and, importantly, fake social-media indicators like purchased followers, likes, or views. Civil penalties run up to roughly $53,088 per violation under the current schedule, and "per violation" can mean per fake review.
• European Union, the Omnibus Directive. Under the EU's modernised consumer-protection rules, a trader who displays reviews must state whether and how it checks that they come from customers who actually bought or used the product, and if it presents them as genuine, it must take reasonable, proportionate steps to verify that. Fake reviews and undisclosed incentivised reviews are prohibited commercial practices. Penalties for widespread infringements reach at least 4% of annual turnover in the member states concerned, or at least €2 million where turnover figures aren't available.
• United Kingdom, the DMCC Act. Since around April 2025, commissioning, publishing, or hosting fake reviews is a banned practice under the Digital Markets, Competition and Consumers Act. The Competition and Markets Authority can act directly, without going to court first, and impose fines of up to 10% of global turnover.

Search engines have their own, separate rules, and it's worth being precise about them, because the internet is full of overstated claims here. Google's review policies prohibit incentivised reviews, review gating (only inviting the happy customers), and coordinated fake engagement, and Google reports removing hundreds of millions of policy-violating reviews in a single year. What that does not mean is "fake reviews destroy your rankings", that's a myth worth retiring. What's actually true is narrower and more useful: spam and manipulation policies can suppress a listing or trigger a manual action, and self-serving review structured data (a shop marking up its own reviews about itself) is not eligible for star snippets in search results. The upside you imagined from faking it mostly isn't there; the downside is now written into law.

Notice how neatly the law tracks the morality. The regulators didn't invent a new wrong, they wrote down an old one. Disclose your connections. Don't buy opinions. Don't hide the negatives. Don't invite only the happy. Every clause is something an honest merchant was already doing.

How to collect reviews honestly, and how to ask well

Here is the good news, and it is genuinely good: the honest path is also the effective one. You do not have to choose between integrity and a healthy review section. You just have to earn it in the right order.

Great service tends to speak for itself. The customers you truly helped will, some of them, tell the world without being asked. But luck sometimes needs a helping hand. Most delighted customers simply forget to leave a review; a polite, well-timed nudge is not manipulation, it's a reminder. The line between a fair nudge and a dishonest one is not blurry at all:

• Tie every review to a real order. Verified-purchase linkage is the single strongest honesty signal you can offer, and post-2024 it's the safest compliance baseline whenever you present reviews as verified, not a nicety.
• Ask everyone, not just the happy ones. Sending review requests only to customers you expect to be pleased is "review gating," and it's specifically prohibited. Invite the whole cohort and take the answer you get.
• Never tie the reward to the sentiment. A small thank-you for leaving any honest review can be acceptable if disclosed; a reward that arrives only for five stars is buying an opinion. Incentives, where used at all, must be disclosed and must be sentiment-blind.
• Time the ask around delivered value. The best moment is shortly after the customer has actually experienced the product, the parcel arrived, the module is installed and running, not the instant they click "pay." Ask too early and you get nothing; ask at the moment of "oh, this is nice," and you get a real, specific review.
• Make it a two-minute job. One clear link, a rating, an optional sentence. Every extra field is a customer lost. Don't script their words with a leading prompt, or you'll manufacture the "repeated wording" red flag yourself.
• Keep the negatives up, and answer them. A visible, well-handled one- or two-star review does more for your credibility than another identical five. It proves the section is real, and it shows the next buyer how you behave when something goes wrong.

The benefits of doing it this way compound, and they're worth naming plainly. You get more reviews, because you actually asked. You get better products, because honest criticism is free product research. You get a review section that survives scrutiny, from a customer, a regulator, or a search engine, because there's nothing in it to hide. And you get the one thing manufactured proof can never buy: a section a careful reader can run the red-flag checklist against and come away more confident, not less.

One more thing, easy to miss: the way you ask has to be as transparent as the reviews themselves. If you offer an incentive, say so, on the review. If you verify purchases, say that too, it's a selling point, not a disclaimer. Transparency about the collection method is part of the honesty, not separate from it. A review section that quietly filters, quietly rewards, or quietly imports without dating things clearly has already stepped off the path, even if every individual review is real. The rule of thumb: if you'd be uncomfortable printing your review-collection process next to the stars, the process is the problem.

You don’t need custom code for any of this. If you collect reviews on your own store, tie each one to a verified order and keep the negatives visible; if you’d rather gather them through a third party, our Google Customer Reviews and Trustpilot integrations bring verified, invitation-based reviews onto your product pages. And since trust is built from more than stars, it’s worth checking that the rest of your store looks trustworthy too. The tool matters less than the rule: verify, don’t gate, and answer the critics.

Sidebar: the same rule applies to likes

A field checklist you can keep

Print this, or keep it in your head. It works on a supplier, a competitor, a module you're about to install, and on yourself. Every item is a question, and every "yes" is a place to go ask for the provenance rather than a place to reach a verdict.

Our own glass house

Writing this article is signing a permanent invitation to be audited, so we should go first. We ran the checklist against our own store, and here is the honest result, not a victory lap.

We are early on reviews, and we're not going to pretend otherwise. We have a small number, not a wall. Some of them we imported from Google and an open review platform, and, here is exactly the caveat we described at the top, those imports carry the date we brought them in, which clusters them together. If you ran the red-flag checklist on us cold, you'd rightly flag that clustering and ask us for provenance. The provenance is: they're real reviews from real platforms, and the cluster is an import date, not a busy afternoon. We're telling you that before you have to ask, because that's the whole point of the article.

We'll also admit the less flattering part: our sample is currently too small to have a meaningful distribution at all, and while every on-site review is now automatically checked against a real order before it counts as verified, we recently removed a leftover demo review rather than let an unverified placeholder sit in a section that's supposed to mean something. By our own checklist, "too few real reviews yet" is a fair thing to flag about us. So we're fixing it in the open rather than dressing it up.

What we are deliberately not doing is manufacturing a tidy five-star chart to decorate this section. We could. It would be trivial. It would also make us exactly what this piece is about. So instead we're doing the slower thing: wiring every on-site review to a verified order, keeping incentives out of it, leaving any critical reviews up and answered, and publishing a permanent Review Integrity page that states plainly how our reviews are collected and verified, so the method in this article can be turned on us at any time.

We would genuinely rather have a few real reviews than a wall of fake ones. A few real ones you can trust. A wall you have to investigate. And in the end that is the entire argument: you cannot fake trust for long, because trust is the one asset that only exists in other people's minds. Real, boring, verifiable social proof, the kind that arrives unevenly, admits the occasional four stars, and can survive a stranger reading it closely. Always wins, because it's the only kind that was ever actually yours.

If you'd like to build that kind of section on your own PrestaShop store, every review tied to a real order, negatives kept honest, nothing gated, that's the philosophy behind our own Product and Store Reviews module. But the method in this article costs nothing and works on any store, including ours. Use it on us. That's the idea.