How to respond to customer reviews without making things worse.

Replying to reviews well depends less on writing skill than on a routine. Sort each review by risk before writing anything; draft the low-risk replies quickly and specifically; run every draft through the same short list of checks; and send the high-risk reviews to the owner rather than to an assistant or a template. A person posts every reply.

The rules below come from what the review platforms publish themselves, in Google's Business Profile help and Maps content policy and on Yelp's support pages, and from the US Federal Trade Commission's rule on consumer reviews, which took effect on 21 October 2024. They are a working method, not legal advice. A review that mentions an injury, a claim or a lawyer is a reason to take advice before replying.

On Google, an approved reply is posted publicly under the review, the reviewer is notified, and the reviewer can still change their review afterwards. Replies are checked against Google's content policies before they appear; Google's help page says this usually takes up to ten minutes and occasionally up to thirty days, and a reply that is not approved has to be edited. A business can edit or delete its reply later, but by then people may already have read it.

That changes what a good reply is. The reviewer already knows what happened. The reader who matters is the next customer, who wants to see how the business behaves when something goes wrong: whether it listens, whether it argues, and whether it would treat them the same way.

Triage decides who writes the reply and how carefully, and it takes seconds. Most reviews are green. The point of the system is to make sure the red ones are never answered from a template at the end of a long day.

A short checklist applied to every reply catches more problems than careful writing does. Each item below is either a platform rule or a direct consequence of the reply being public.

• NO PRIVATE DETAILS

  Never confirm the customer's address, what was found in their home, what they paid or anything else they did not put in the review themselves. Google's reply guidance says plainly never to share a reviewer's private information.

• NO ARGUMENT

  Disagree with a fact once, briefly and politely, if it matters to future customers. Never argue about the reviewer's experience.

• NO INCENTIVES

  Never offer anything for a review, for changing one or for removing one. Google's Maps policy prohibits incentives in exchange for a review or for removing a negative one, and Yelp prohibits offering freebies, discounts or payment for reviews or for their removal.

• NO INVENTED FACTS

  If you do not know what happened on the job, do not describe it. Check the job record first, or say that you are looking into it.

• SPECIFIC, NOT COPIED

  Mention something the reviewer actually said. Google's tips advise against sending the same thank-you to everyone.

• SHORT

  A few sentences. Google's guidance is to keep replies short and simple, because long ones lose their readers.

• A PRIVATE ROUTE

  For anything amber or red, give one clear way to continue the conversation privately, and then use it.

The shape that works is short. Thank the reviewer for raising it, acknowledge the experience without conceding facts you have not checked, correct one fact only if leaving it would mislead future customers, and give a named way to talk privately. Then make the private follow-up happen, because the public reply is a promise that it will.

What fails is equally consistent: a long defense, a timeline of the job, a suggestion that the reviewer is lying, or a hint that a refund depends on the review coming down. The FTC's rule prohibits using threats, intimidation or false accusations to get a negative review removed, and the FTC's guidance specifically mentions unfounded legal threats. Contacting a customer to put a problem right is not suppression under that guidance. Offering something in return for the review coming down breaks Google's and Yelp's incentive rules.

Asking a customer, once a problem has genuinely been put right, whether they would consider updating their review is treated differently by different rules. The FTC's guidance says that asking satisfied customers to update a review is not suppression. Yelp's support pages tell businesses not to ask anyone for reviews at all. Where the rules differ, follow the strictest one that applies to the platform the review is on.

Some reviews are not from customers. Replying with an accusation is risky twice over: if you are wrong, you have publicly called a real customer a liar, and the FTC rule prohibits false accusations about a reviewer made knowingly or recklessly. Where a public reply is needed at all, the safer one states only what you can verify, that you cannot find a record of the job, and invites the reviewer to get in touch. The real response happens through the platform's reporting process.

Extortion is a separate case. Google describes a pattern of sudden one- and two-star reviews followed by a demand for payment or favors to remove them. Its advice is not to pay or negotiate, to preserve the evidence, including screenshots of the demand with dates and sender details and links to the reviews, and to report it through its dedicated extortion form. That is an owner's job rather than an assistant's, and it is a red review by definition.

Replying and asking are separate activities with separate rules, and those rules diverge. Google allows a business to ask customers for genuine reviews, including through its own review link, but prohibits incentives and prohibits selectively asking only the customers expected to be happy. Yelp tells businesses not to ask anyone for reviews, not to offer incentives, and not to ask for reviews after collecting feedback elsewhere, such as through a survey.

The FTC rule sits underneath both. It exempts general requests made to all customers, does not prohibit incentives that are unconditional on what the review says, and prohibits paying or rewarding a review for expressing a particular sentiment. The FTC's guidance adds that asking only the customers you believe are happy is not specifically banned by the rule but could still break the wider FTC Act. A business that asks every customer the same way, offers nothing for it and follows the strictest platform rule is on the defensible side of all three.

A chat AI is good at the first draft of a green reply and at checking any draft against a list. Give it a short description of the business's voice, the checks above and the text of the review, and it will produce a reasonable reply in seconds. Give it a second, separate instruction to check a draft for private details, incentives, argument and invented facts, and it becomes a useful second reader.

Three boundaries keep it safe. It drafts only from what the review says and what you tell it, never from guesses about the job. It does not handle red reviews. And it does not post: nothing should log into the business's review accounts on its own, and a person reads every reply before it goes out. The time saved is in the drafting, not in the judgment.

Platform rules come first, and they are not all the same. Yelp's guidelines tell users not to use third-party AI tools and chatbots to create reviews or other content, including to draft or revise it, and its support pages say AI-generated reviews break its Content Guidelines. Its pages for business owners on responding to reviews do not mention AI, and Yelp does not say whether the general rule covers an owner's response. Until it does, write Yelp responses yourself rather than from an AI draft, and check each platform's current rules before using AI drafts anywhere else.

Reviews answered in a weekly batch get more consistent replies than reviews answered whenever someone remembers. Set a time, work through new reviews by triage level, log what was posted, and clear a few old unanswered reviews each week until the backlog is gone. A late reply to an old review still shows the next reader that the business pays attention.

Once a month, read the month's reviews together and look for the theme: the same complaint about arrival windows, the same praise for one technician's explanations. Pick one operational change to try. Reviews are the cheapest customer research a service business receives, and a reply routine that never feeds back into the work misses most of their value.

Aevrion Ops builds review workflows into a business's own systems, covering triage, drafting, checks and routing to the right person, where each platform's terms and API allow it. Posting, and every decision about a red review, stays with the business.

1. Write for the next customer reading the reply, not only for the reviewer.
2. Triage green, amber and red before drafting; red reviews go to the owner, never to a template.
3. Run every draft through the same checks: no private details, no argument, no incentives, no invented facts, specific and short.
4. Acknowledge, correct at most one fact, move the conversation offline, and then follow up privately.
5. Report suspected fakes and extortion through the platform; never accuse publicly and never pay.
6. Asking for reviews has different rules on Google, on Yelp and under the FTC rule; follow the strictest one that applies.
7. Let an AI draft and check replies where the platform's rules allow it, keep red reviews and every post with a person, and write Yelp responses yourself while Yelp's position on AI-assisted owner responses is unstated.