Why asking only your happy customers for reviews backfires
Filtering review requests so only satisfied customers get asked is called review gating. Google prohibits it explicitly, and the other major review platforms carry their own version of the same rule. It is also, separately from the rules, a bad trade: it buys a rating that customers have learned to distrust.
Published by Rumore.
The pitch is easy to understand. Survey the customer first, and only send the review link to the ones who answer well. The average goes up, the bad reviews go somewhere private, and nothing was faked because every review is from a real customer who really was happy.
Google's review policy forbids discouraging or prohibiting negative reviews, and forbids selectively soliciting positive ones. The published penalties run from having the affected reviews removed to having the listing itself suppressed.
Set the rules aside for a moment, because the practical argument is the stronger one. Think about how you read reviews yourself: the bad ones tend to get opened first, and what you are really checking is whether the complaints are things you would mind. A business with hundreds of reviews and nothing below four stars invites the question of what happened to the rest. A few mediocre reviews with thoughtful replies underneath them are more persuasive than an unbroken wall of five stars, because they are legible as real.
There is also what gating costs you internally. The unhappy customers are the ones telling you which thing is broken, and routing them into a private form that nobody reads on Monday morning is a way of not finding out. The complaint that arrives publicly is unpleasant and useful. The one that arrives in a suppression funnel is only unpleasant.
The version that works is duller: ask everyone, ask soon after the visit while it is fresh, make the link one tap, and reply to whatever comes back. That produces a rating that moves slowly and holds up.
The line is not about asking, it is about who gets asked. A satisfaction question is harmless on its own. A satisfaction question that decides who sees the public link is the entire practice. So is a card handed out only on the days that went well, and so is a staff instruction to mention it to the pleased ones. None of this depends on software. A business can gate with a clipboard.
Incentives are a separate rule that gets broken at the same counter. Offering a discount, a free coffee or an entry into a draw in exchange for a review is prohibited whether the review turns out positive or not, because paying for reviews distorts the sample in the same direction as filtering does. Asking is free and allowed. Buying is neither.
What the published rule does not say is how a platform decides which reviews were affected, and that is the part worth worrying about: you do not get to choose, and you will not be shown the working. A rating built over years is not something you want resolved by a process you cannot see.
If you want one test for whether something you are already doing counts, use this: would you be comfortable showing a customer the rule you are applying? "We ask everyone the day after their appointment" survives that. "We ask you because you seemed happy" does not, and neither does any version of it dressed up as a survey.
Does this count as review gating?
- A survey whose answer decides who gets the review link
- Yes. The branch is the practice, whatever the survey is called internally.
- A survey everyone gets, alongside the same link
- No. Nothing is being sorted, so the survey is research.
- Handing out the card only on the days that went well
- Yes. Nothing about gating depends on software. The filter is the decision, not the tool.
- "Tell us first if something went wrong", with no link for the unhappy
- Yes, if the unhappy never see the link at all. The wording is not what decides it. The two paths are.
- Offering a discount or a free item for a review
- A different rule, and Google prohibits it outright. Nothing about it depends on the review turning out well.
- Asking every customer, the same way, every time
- No. This is the thing the platforms expect you to do.
Common questions
Is it against the rules to ask customers for reviews at all?
No. Asking every customer is allowed, and is what the platforms expect. What is prohibited is selecting who gets asked based on how satisfied they are, or offering something in exchange for a review.
Can you ask a customer to contact you before leaving a bad review?
A request framed as "tell us first if something went wrong" is still gating if it only routes unhappy customers away from the review link. Offering everyone both (a review link and a direct line) is fine. Using one to divert people from the other is not.
Is a satisfaction survey before a review request always gating?
Not by itself. A survey becomes gating when its answer decides who gets the public link. Send the survey and the link to everyone, in the same message or the same sequence, and the survey is research. Use the answer to sort people down two different paths and it is the practice, whatever it is called internally.
Can you offer a discount or a free item for a review?
No. Google prohibits offering anything of value in exchange for a review, and the major review platforms carry their own version of the rule. It is separate from gating and it is a separate way to lose the reviews you already have. You can ask as often as you like. You cannot pay.
What actually happens to a business caught filtering reviews?
Google publishes two consequences for the listing: the affected reviews come off, and beyond that the listing itself can be suppressed in search. What it does not publish is where the boundary of "affected" falls, which is the whole problem. The risk is not a fine you can budget for. It is a rating you cannot rebuild on demand.
Is it gating to ask in person only when the visit went well?
Yes. Deciding at the counter who gets handed the card is the same filter as a branch in an email sequence, and it produces the same distorted sample. Ask everybody, in the same way, or do not ask.