How to Respond to a Negative Google Review as a Vet: AI-Drafted Replies

A vet clinic can use AI to draft a reply to a negative Google review if the reply never confirms that the reviewer is a client, never mentions the animal or the visit, and invites the person to contact the clinic privately. Give AI only the public review text and fixed rules, then have a person approve it.

The trap is kindness. A warm reply that says "we are so sorry Bella's treatment did not go as hoped" has just confirmed the client, the pet and the fact that there was a treatment, in public, to anyone searching your clinic. The safe version has to be the easy one, so every pattern below starts from what a reply must never say and only then adds the warmth.

Follow me on Instagram@sagnikteaches

Why a vet's reply is different from a café's

A restaurant can write "sorry your steak was overcooked". A vet clinic holds clinical records, and veterinary codes of conduct treat information about a client and their animal as confidential: the codes I checked say it must not be disclosed without the client's permission, except where the law or an animal's welfare requires it. One regulator's guidance on social media goes further and addresses this exact situation, telling vets to avoid disputes in a public forum, to invite the person to contact the practice directly, and to use discretion about how much to say publicly. Replying to a review is not an exception, and a reviewer posting publicly is not the same as giving you permission to discuss their case. The practical consequences are:

Connect on LinkedInSagnik Bhattacharya

• You cannot confirm the reviewer is a client, even if the review makes it obvious.
• You cannot correct a factual error about the case, even when the reviewer is plainly wrong.
• You cannot mention fees, treatments, outcomes, dates or staff involved in a specific case.
• You can state general facts about how the clinic works, apologise that someone feels let down without saying why, and offer a private route.

That leaves you with less to say than most owners want. The skill is saying it well, quickly, and in a way that readers who have never met you see as calm. Ask your professional body or indemnity insurer for their current guidance on online reviews; what follows is a practical routine, not legal advice.

Subscribe on YouTube@sagnikteaches

What to give the AI and what to keep out

The review is already public, so pasting its text into an AI tool adds little risk. The risk is what you add from your own side. A simple rule: the AI sees the review and the rules below, nothing else.

• Allowed in: the public review text, the star rating, your clinic's general policies (opening hours, how complaints are handled, the phone number or email you want people to use).
• Never in: the clinical record, invoices, appointment details, client names from your system, any note about what happened, even a short "she came in on the 14th".

Use a business or team plan of the assistant, where content is not used for training by default (ChatGPT Business and Claude Team both list at $20 a seat a month annually, $25 monthly, with a two-seat minimum), or an individual plan with training switched off. If you do check the record to understand the complaint, do it in your own system, in your own head, before you start the prompt.

Five reply patterns that cover most negative reviews

Almost every negative vet review falls into one of five shapes. Naming the pattern first makes the reply faster and keeps the AI inside the lines.

Pattern 1: The grief review

Someone lost a pet and blames the clinic. This is the most sensitive and the one where a fast, generic AI reply does most damage. Reply with brief acknowledgement, no defence, an offer to talk, and nothing else.

Thank you for taking the time to write. We are sorry to read how distressing this has been. We would like to listen properly and would be grateful if you would contact our practice manager on [phone] or [email], so we can give your concerns the time they deserve.

It does not say "your pet", "our loss", "treatment" or "sorry for your loss" (all of which confirm a case). It names a person to contact. Any review alleging negligence or wrongful death goes through the escalation rule below before this is posted.

Pattern 2: The fees review

"Charged me a fortune for a simple visit." Do not discuss this person's bill. Do describe your general approach, which is useful to future readers.

We are sorry to hear this. We always aim to explain the options and costs before treatment, and we would like to hear more about your experience. Please contact [name] on [phone] or [email] so we can look into your concerns.

This works only if it is true. If your clinic does not give estimates, the sentence must change, not the facts. The tutorial on payment plan policy wording helps make fee explanations part of your normal process, so the review does not arise.

Pattern 3: The service review

Rude phone manner, a long wait, no call back. These are the reviews you can learn from and that carry no clinical detail.

Thank you for your feedback. We want everyone who contacts us to feel listened to, and we are sorry if that was not your experience. We would welcome the chance to look into this: please email [address] with your contact details and we will respond personally.

Notice that it does not say "our receptionist was having a busy morning" or any explanation of what happened, because any explanation implies you know who the person is.

Pattern 4: The vague one-star

One star, no text, or "Awful." There is nothing to answer, so keep it brief and neutral.

Thank you for your feedback. We are always keen to improve, and if you would like to tell us more, please contact us on [phone] or [email].

Pattern 5: The review you suspect is not genuine

Possible competitor, someone who never visited, or a review that describes an event that does not match anything you know. Your instinct is to say "we have no record of you". Do not: confirming that someone is not a client is as much a disclosure as confirming that they are, and it risks being wrong.

We take all feedback seriously, but we cannot discuss any individual's contact with the clinic in a public forum. If you would like to talk this through, please contact us on [phone] or [email]. We have also asked the platform to review this post.

Then report the review from your Business Profile: next to it select Report, choose the reason and send. Google's help page says evaluation typically takes several days, that only reviews breaking its policies are removed (not ones you simply disagree with), and that you can check the status and make one appeal in its Reviews Management Tool. Google also checks each reply you post against its content policies; that usually takes up to 10 minutes but can take up to 30 days, and you can edit or delete a reply later from the Read reviews page.

The prompt that produces a safe draft

You are drafting a public reply to a Google review for a veterinary clinic.
Rules, all mandatory:
1. Never say or imply that the reviewer is a client or that an animal was treated.
2. Do not mention a pet, a visit, a treatment, a fee, a date or a member of staff.
3. Do not defend, explain, correct or argue. No blame of any kind.
4. Do not offer refunds, compensation or admissions of fault.
5. Apologise only that the person feels let down; do not say what happened.
6. Invite private contact using exactly this: [phone] or [email].
7. 50 to 80 words, calm, plain English, no exclamation marks, no emoji.
8. Start with "Thank you" and end with the contact line.

Classify the review as one of: grief, fees, service, vague, suspected fake.
Show the class first, then the draft, then list any detail in the review that
you deliberately did not respond to.

Review (star rating then text):
[paste]

The last instruction is the useful one. When the AI lists what it deliberately ignored, you can see whether it understood the reply must not engage with specifics.

A worked example, with the fixes

An illustrative review, one star: "Took my dog in on Saturday for a limp. They did blood tests I didn't ask for, charged more than $300, and the vet barely spoke to me. Dog was no better. Never again."

Illustrative first draft from an AI with no rules: "Thank you for your review. We are sorry that your dog's limp was not resolved. Blood tests are a standard part of diagnosing lameness, and our vets always explain them. We would be happy to discuss the invoice." What is wrong, line by line:

• "your dog's limp" confirms a client, a pet and a condition.
• "Blood tests are a standard part of diagnosing lameness" is clinical defence of a specific case, and arguable.
• "our vets always explain them" contradicts the reviewer in public and is the kind of absolute that gets quoted back.
• "the invoice" confirms there was one.

Using the prompt above, the class comes back as "fees plus service", and the draft is the Pattern 2 wording with the service line added: "Thank you for your feedback. We are sorry you feel let down. We aim to explain options and costs clearly and to give every person time, and we would like to hear more. Please contact [practice manager] on [phone] or [email] so we can look into your concerns." It ignores all the detail, which is what you want. It still needs one human edit: "we aim to explain options and costs clearly" should be there only if it is a real clinic standard.

The escalation rule: when AI does not draft at all

Some reviews should not be answered by anyone until a named person has read them. Write this rule down and pin it where the reviews are read. A review is red if it contains any of these:

• An allegation of negligence, malpractice or a death caused by the clinic.
• A threat of legal action, a complaint to a regulator, or a media mention.
• A named member of staff, or a claim about an individual's conduct.
• Personal threats, harassment or discriminatory language.
• Details that suggest an unresolved formal complaint already in progress.

For a red review: no AI draft goes online. The practice owner or clinical director reads it the same day, considers contacting the clinic's indemnity insurer or professional body adviser before any public reply, and decides between a minimal Pattern 1 reply, no reply and a private approach first. An amber review mentions fees, wait or manner with no allegation: AI drafts, the practice manager approves within two working days. A green review is positive or neutral: AI drafts a thank-you, and anyone on the team can approve it.

Who approves is as important as what is written. The person who wrote or handled the case should not be the sole author of the reply, because they will want to defend themselves. Use a second pair of eyes, ideally the practice manager.

Positive reviews, and the same trap in reverse

The same confidentiality applies when a review is glowing. "Thank you for trusting us with Max's surgery" confirms a client, a pet and an operation, even though the person volunteered it. Safe positive replies stay general: "Thank you for the kind words, it means a lot to the whole team. We are always glad to hear when someone has had a good experience." A short before and after: Before, "So pleased Max is recovering well after his op!" After, "Thank you so much for taking the time to write this; the team will be delighted to read it."

The discussion of auto-posting replies to Google reviews explains why an approval step is worth keeping even for positive ones; for a vet clinic, it is not optional.

Learning from reviews without breaching anything

Reviews are free market research. Once a month, copy the public text of the last 20 reviews, with names removed, into an AI chat and ask for recurring themes: wait times, phone access, fee surprises, manner. The method for analysing reviews with AI works unchanged. Then fix the process. A recurring "nobody called me back" review is a reason to look at call routing, and what an AI receptionist costs and when calls go to staff covers how to route urgent calls to people.

A small illustrative sum: if 6 of the last 20 reviews mention not being able to get through on the phone, that is 30 per cent. Cutting that with a call-back promise or a second line is worth more than ten well-worded replies.

Two more reviews, walked through

Practice on invented reviews before a real one arrives. These two are illustrative and show how the classification changes the reply.

Review A (two stars): "Phoned four times before anyone picked up and then was told to ring back after lunch. Hopeless." This is amber, Pattern 3. It names no case, no person and no outcome. The draft is the service wording, and the follow-up is internal: check the phone log for that morning and ask whether lunchtime cover is a gap. The review is a gift to the process, even if the tone is not.

Review B (one star): "The vet was negligent and my cat died because of them. I will be reporting this." This is red on two counts, an allegation of negligence and a threat to report. No AI draft goes anywhere near it. The practice owner reads it, contacts the indemnity insurer for advice, and decides whether to post a minimal Pattern 1 reply after that advice, or no reply at all while a private approach is made. The cost of waiting a day is small; the cost of a wrong public sentence on a possible formal complaint is large.

A sharper test is to run both through your prompt and see whether the classification matches yours. If the AI calls Review B "grief" and drafts a sympathy reply, the escalation rule is doing its job by existing outside the AI: the red check happens before any prompt is run, by a person.

A ten-minute drill for the team

Staff who have never written one of these replies will reach for natural sympathy. A short drill fixes that. Print five invented reviews, one per pattern. Ask each person to draft a reply by hand in three minutes, then compare with the AI draft. Mark every sentence that confirms a client, mentions an animal or defends the clinic. Most first attempts have two or three. Repeat once a quarter, using the latest real wording (names removed), so new starters learn the habit and nobody drifts back to kindness that breaches.

It also helps to keep three or four approved clinic facts that the AI may use, written as true statements: "We aim to explain costs before treatment", "We answer complaints within [number] working days", "Our practice manager is [role]". When every draft draws on the same short list, claims stay accurate and replies stay consistent without sounding copied.

A short script for the practice manager

1. Read the review twice. Check the date and star rating, and look up the clinic record privately if you need to understand the complaint. Do not copy anything from it.
2. Classify it: red, amber or green; and pattern 1 to 5. Red goes to the owner now.
3. Run the prompt. Paste only the public review.
4. Check the draft against the rules: no pet, no visit, no fee, no staff, no defence, no admission, a private route given.
5. Post it, then note it. Log the date, the class and who approved it. If Google does not approve the reply, it asks you to edit it; you can also change it later from the Read reviews page. The reply shows as from the business, not from your personal name, and the reviewer is notified, so expect some to edit their review after reading it, which also changes the date shown on the review.
6. Follow up privately. If the person contacts you, the real conversation starts offline, with the record in front of you and a clinician in the room where appropriate.

The log in step 5 has a second use. If you can show that every negative review was read within a day and answered within two, that is evidence of a working process when a partner, insurer or auditor asks.

Where AI drafts go wrong in this job

• Sympathy that confirms. "Sorry for your loss" or "sorry about your cat" in a first draft. Scan every draft for animal words.
• Policy claims you cannot back. "We always provide a written estimate" when you do not. Tell the AI the three true things about your clinic and let it use only those.
• Defensive tone creeping in. Phrases like "as we explained at the time" or "our records show". Ban them in the prompt.
• The same reply under every review. Readers see three near-identical paragraphs and trust you less. Vary the opening and the closing, while keeping the safe core.
• Posting under pressure. A bad review at 6pm on a Friday feels urgent. Unless it is red, it can wait until Monday morning, and a calm reply two days late beats a fast one that breaches confidentiality.

Questions vet owners ask about replying to bad reviews

Can I say the reviewer's pet was treated well?

No. Even a defensive-sounding reassurance confirms the person is a client and implies a case. Keep to what you do in general, not what happened to this animal, and move the conversation to a private channel.

Should I reply to every negative review?

Reply to the ones a future client will read and judge you by, within about two working days. A short, calm reply to a rude one-star with no detail is still worth posting, as long as it says nothing about whether the person used the clinic.

What if the review is fake or from a competitor?

Report it from your Business Profile (select Report next to the review and pick the reason), keep the public reply neutral and short, and do not accuse anyone. You cannot prove the reviewer was not a client without checking records, and confirming that publicly would breach confidentiality either way.

Is it safe to paste a review into an AI tool?

The review text is public, so pasting it is low risk if you leave out your own records. Never paste notes, invoices, client names or anything from the patient record, and use a business plan where content is not used for training.

Further reads

• Writing Pet Owner Emails and Treatment Explanations With AI — Pet owner emails and treatment explanations done safely.
• AI Assistant for Veterinarians: Scribe, Phone or Chatbot First? — Which AI tool to start with in a vet practice.
• Can AI Handle Vaccination and Check-Up Reminders for a Vet? — Routine reminders that cut avoidable complaints.
• How to Answer Restaurant Complaints With AI Without Escalating — A complaint-reply method from hospitality, adaptable here.
• AI Receptionist for a Veterinary Practice: Setup and Call Scripts — Call scripts that keep urgent calls with people.
• How to Reply to Hotel Reviews With AI Without Sounding Canned — Reply wording that avoids sounding canned.
• How to Respond to Negative Roofing Reviews With AI: Claims and Delays — Eight example replies to negative roofing reviews about claims and delays, plus an AI prompt that drafts them without leaking claim details.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Google Business Profile help pages on replying to reviews and on reporting inappropriate reviews; veterinary professional conduct guidance on client confidentiality and social media; ChatGPT Business and Claude pricing pages; the reply examples are illustrative.