There is exactly one shortcut in review automation and it is the one you must not take. Review gating, sending the public review link only to customers you expect to be happy, is against Google's review policies and, in the United States, sits squarely inside what the Federal Trade Commission's rule on consumer reviews prohibits. Plenty of platforms still ship it as a feature, usually described as a satisfaction check or a feedback filter. Turn it off.
Everything else about this is timing, wording, and making sure somebody asks at all.
Ask everyone, or ask nobody
The rule is that simple, and it is also the one that makes the system work.
A business that asks every completed customer gets a review profile that looks like its actual service. That includes the occasional bad one, which is fine and arguably necessary: a profile with nothing but five-star reviews reads as bought, and customers discount it. A business that asks selectively gets a rating that cannot survive contact with a bad month, because the moment the filtering stops the average falls off a cliff and the recent reviews all disagree with the old ones.
There is a practical argument too. Selective asking means somebody has to decide who is happy, which means somebody has to remember, which means the process stops being automatic and starts being a job nobody does.
The trigger is most of the timing problem
Pick a completion event that genuinely proves the customer experienced the service. Work order marked complete by the technician, delivery confirmed, invoice paid, appointment attended. Not "lead created", not "job scheduled", and not a date on a calendar.
Then decide the delay deliberately. Too fast and you are asking before the customer knows whether the repair held. Too slow and the job is out of mind and the technician's name is forgotten. For same-day work, later the same day or the next morning is usually right, once the customer has lived with the result for a few hours. For longer projects, wait until the final walkthrough is genuinely done and any snags are closed, because a request that arrives while an unresolved complaint is open will get answered, and not in the way you wanted.
One reminder, a few days later, to people who did not respond. Not a sequence.
Write it so it reads like the person who did the work
The best review requests are short, specific, and obviously from a human being.
Name the business and, where you can, the technician who was there. Reference the job in a few words so it is clear this is not a bulk send. Ask once, plainly, and make the request optional in tone as well as in fact. One link, going straight to the review form rather than to your profile page, because every extra tap costs you completions.
What must not appear: a suggested star rating, suggested wording, a sample review, any incentive, or the phrase "if you were happy with our service". That last one is soft gating and it does the same damage. Do not offer a discount, a prize draw, or a gift card in exchange for a review. It breaks Google's policy, it breaks the FTC rule if the review is not disclosed as incentivised, and it produces reviews that read like they were paid for, which they were.
SMS generally outperforms email for trades, because the phone is where the customer already is. If you go that route the same carrier rules apply as for any other business texting: registered messaging, working opt-outs, and a message that identifies you in the first few words. Our guide to missed-call text-back automation covers that ground in more detail.
What you cannot do, stated plainly
- Ask only customers you expect to rate you well.
- Offer anything of value in exchange for a review.
- Write reviews for customers, or have staff, family, or contractors post as customers.
- Buy reviews, or buy a service that generates them.
- Suppress, hide, or bury a genuine negative review, or threaten a customer who posts one.
- Set up a feedback form that intercepts unhappy customers before they reach the public link.
The FTC rule carries real penalties and it applies to the business, not to the software vendor who suggested the workflow. "The platform had it switched on by default" is not a defence anyone will accept.
Negative feedback needs a route, but not instead of the link
There is a legitimate version of routing unhappy customers to a person, and it is easy to tell apart from gating.
The legitimate version: every customer gets the same message with the same public review link, and the message also offers a direct way to reach the office if something went wrong. Both paths are open to everybody. Somebody with a complaint frequently prefers the direct route, which is good for them and good for you, and they can still leave a public review if they want to.
The version that is gating: the first message asks how it went, and only the positive answers see the review link. Same intention, different mechanism, still prohibited.
If your platform cannot do the first version, that is a reason to change platform.
Respond to what arrives, and treat the pattern as information
Reply to reviews, including the good ones, in a voice that sounds like your business. Keep the negative replies short, factual, and free of the customer's private details, which means no case numbers, no addresses, no discussion of what they paid. Offer the conversation offline and then actually have it.
The more valuable use of reviews is diagnostic. Three months of them will tell you where the operation leaks: the same complaint about arrival windows, the same praise for one technician, the same confusion about what the quote covered. That is free operational research and almost nobody reads it as such.
AI can draft responses and it is decent at it, but keep a person on anything sensitive, specific, or potentially private. A generated reply to a serious complaint reads exactly like what it is.
Volume and recency, not just the average
A high average built from a handful of old reviews is weaker than a slightly lower average with a steady flow of recent ones. Recency signals that the business is still operating and still being chosen, and it is the thing a steady request habit produces almost automatically.
That is the real case for automating this. Not that automation writes better requests, but that it makes the asking consistent. Most local businesses do not have a review problem, they have an asking problem: somebody remembers during a quiet week and forgets for the next two months.
What to measure
Requests sent is not the number. Track the chain: requests delivered rather than sent, click-through to the review form, reviews actually published, and the gap between the last two, which is where a clumsy link or a login wall is costing you. Watch complaints and opt-outs as your signal that the timing or wording is off.
Then watch the outcomes that matter commercially: review count and recency over time, and what happens to the calls you get from the map pack. Reviews are one of the inputs to local ranking, though not in the simple way most people assume, and our piece on why a profile ranks in one part of town and not another explains what else is moving underneath.
Where it sits
Review generation is one habit inside profile management rather than a product on its own, and it works best when the profile it feeds is properly set up and actively maintained. Our Business Profile setup guide covers the foundation, and what profile management costs separates the one-off cleanup from the ongoing work. If you want the whole thing run rather than assembled, that is Google Business Profile management.
The habit is what matters. A short honest message, sent to every finished customer, at the right moment, forever.
FAQs
Can we ask only happy customers?
No. Filtering based on expected sentiment can violate platform policies and distorts feedback.
How many reminders should we send?
Usually one timely request and a restrained follow-up are safer than a long sequence.
Should AI write review responses?
AI can draft responses, but a person should review anything sensitive, specific, or potentially private.