The Google review timing rule clinics keep getting wrong

The visit that earns a Google ask is already over. Checkout closed. They left the room. Nobody is standing over the card reader. The thread does not contain a complaint about the wait, the bill, or the table.

The visit that must never get the text is Bao, still in the parking lot, writing that he sat twenty-five minutes and the room was cold. That thread is a dispute. A person takes it. Sending the Google link into it is how a clinic trains people to associate the brand with a fight, and it is also how you wander toward the kind of selective solicitation Google and the FTC both dislike.

The honest ask is not a drip, a tablet in the treatment room, or a prompt that says “hope the injection helped.” It is one genuine ask after a completed visit, in the language already on the thread, pointing at the Google Business Profile for that address. If your desk already does that by hand, you do not have this hole.

Ask after checkout, not on the table

Google’s merchant rules are the first constraint, not a vendor feature. Merchants may solicit a genuine review. They may not pay for stars, tell people what to write, request specific content including a staff name, or pressure a rating on the premises. The FTC’s fake-reviews rule, announced 14 August 2024 and effective 21 October 2024, is the same shape on the legal side: fake reviews, buying a particular sentiment, and certain kinds of suppression are out.

“On the premises” is the part clinics miss. Yara Mensah used to catch people at the desk. Then the book filled. Someone had the idea to ask while they were still on the table, or while Dr. Malik Rahman was in the doorway with the next chart. That is pressure. The patient is half-dressed, they want to leave, and the person who just treated them is watching. The recap at the end of the visit is still treatment. “How does the shoulder feel” is a clinical conversation, not a review window.

Ask after they have left the room and checkout has closed. A text on the way to the car is the usual shape. A quiet sentence at the door after they are dressed, without watching them type, is closer to allowed than a tablet in the room. A card-reader prompt under a practitioner’s eye is the shape Google is describing. If you are going to staff software, prefer the text after they have gone. It is easier to stop, easier to keep clinical language out, and easier to point at the right listing.

“Good visit” here means the visit is done on the book. They showed. Duration finished. Checkout closed. No open complaint on that thread. It is not a smile score Malik typed, and it is not a model that skips quiet people. Corinne said almost nothing at the door. She still gets the same one ask as everyone else whose visit closed cleanly. Skipping her because she seemed unlikely to be pleased is selective solicitation. Stopping Bao’s send because he already wrote a complaint is a dispute handoff.

The consumer-floor version of the same job, with different nouns, is How to ask for a Google review after a salon visit. Do not copy a service-name prompt onto a clinic thread.

Do not ask for a clinical outcome

The SMS can say the visit is done. It cannot name a diagnosis, a treatment, a body part, or a room note. “If you have a minute, a Google review helps other patients find the clinic” is the ceiling. “Hope the knee injection went well, please mention Malik” fails three rules at once: clinical leakage, coached copy, and a staff name.

This is HIPAA-aware common sense even before a lawyer walks in. The package field in Zenoti or Mindbody is often “lumbar follow-up 45.” If the review tool pastes that string into SMS, you have put treatment detail on a phone that will be left on a counter. The chart stays in the chart. The public review, if they write one, is their speech. You do not draft it, and you do not ask them to report whether they feel better.

They may volunteer an outcome in the review itself. That is their choice. Your prompt must not invite it. “Did the visit help” is a clinical question wearing a marketing hat. So is a follow-up text that says “if you are feeling better, tell Google.” If they later write a 2-star, you still do not send “please update your review.” That is the suppression shape the FTC rule is aimed at. Yara answers the listing as a person, or she does not answer.

Incentives are out. A discount, a gift card, a credit for a star. One ask means one ask, not a Sunday nudge, not a card-reader plus email plus text. Leona already left a review last month after a visit with Dr. Elise Navarro. Today’s 45 with Malik does not earn a second ping. Already reviewed is a stop, the same way a named family hold is a stop.

One listing per address

Google Business Profile is per street address. Riverside is one listing. Midtown is another. A two-door group that sends every ask to a single brand page is training Google, and patients, on the wrong door. The person who parked at Riverside and waits for Midtown’s hours to show up in Maps is the person who writes the 2-star you then have to answer.

If you only have one address, this is simple: one short link, used after every completed visit that did not carry a complaint. If you have two, the send has to know which door they sat in. Name, address, phone, and hours on that Google Business Profile have to match the door they actually visited. Sending a Riverside patient a Midtown Maps link is how you collect reviews on the wrong listing and then wonder why Riverside still looks empty in search.

Occupancy across doors is a different job, written up at Who is available across two clinic locations. Do not buy a floor view to fix the ask. Do use the same rule the rest of the group already needs: books stay local. Hours, menus, and listings do not get flattened unless you ask.

A hospital system or an EHR project is not this product. We do not own the chart. We do not write fake replies to a 2-star, and we do not argue with one. Yara still answers Google as a person when a review lands.

How to run the ask this week

You can run this by hand on a short desk. You staff an agent when the book has filled and the ask died.

1. Wait until checkout is closed and they have left the room. Not during treatment. Not at the card reader while a practitioner is watching.
2. Look at the thread. A complaint about wait, room, bill, or care is a stop. A person takes it. Do not send the link into that conversation.
3. Send one SMS in the language already on the thread, with the Google listing for the address they visited. Leave out the diagnosis, the procedure name, a request to mention Malik, and any discount.
4. Do not send a second nudge. Do not also fire the POS drip. One ask across the relationship, not one per practitioner.
5. If they reply with a problem, stop. The desk owns the thread. If they write a review, a person replies on Google.

Read the complaint row next to the same-chance row. They are not the same rule. Never-after-a-complaint cannot mean only-text-the-people-you-expect-to-be-pleased.

Mindbody, Zenoti, Square, and the other books will sell a review toggle. Keep the calendar. If that toggle already sends one ask, in the right language, without the package name, and you can kill it when Bao texts the desk, you may already be done. What the add-on almost never is: a stop on a live complaint in the same inbox Yara is watching, with no clinical leakage, pointed at the correct listing. Front desk is also a different job. It answers while every room is full. Do not staff reputation and expect it to take the lunch line. That write-up is Why "anyone available" does not work in a two-room clinic.

An answering service will not do this job at all. It covers a ring. It does not know Malik’s 45 closed at 9:12. The split is An answering service takes a message. A clinic needs a booked room..

What LLDTEK staffs after the visit

LLDTEK is an AI agent platform. Front desk, recovery, rebooking, reputation, and floor sit beside the POS you already run. Reputation is the ask. Tools: completed visit, language, send once, do not send. Stop: complaint, VIP, already reviewed, checkout still open, anything that would require clinical content to sound “personal.”

You start with the hole. On a growing clinic that hole is often the ask that died when the rooms filled. You do not start with five agents. You watch the run. If it asked after a complaint, it is not live. If it asked in English after a Spanish checkout, it is not live. If it pasted a treatment name, it is not live. If it sent a second nudge on Sunday, it is not live.

Yara still owns the desk. The agent covers the ask she cannot sit still for. Named families, owner friends, and anyone the floor already treats as a person-only thread stay off the send list. A solo practitioner who already asks at checkout does not have this hole. A card on the counter and a sentence at the door, after they are dressed, will beat any agent.

Bring last week’s completed visits to the demo, the complaint that should have blocked an ask, the two languages at the desk, and both Google listings if you have two addresses. What to test on an AI receptionist demo is the longer list. If we cannot show the stop, do not buy the job.