How to Respond to Negative Reviews: Agency Playbook

Respond to a negative review within 24 to 48 hours, in public, in three or four sentences: acknowledge the specific thing that went wrong, state what you're doing about it, and hand off to a named person offline. Don't argue the facts, don't reveal anything about the customer's account, and don't let the client's owner post while angry. For an agency managing this across a book of service business clients, the wording is the easy half — the hard half is the approval loop, because a response that takes nine days to clear is worse than a mediocre one that went out on day two.

Last updated: September 2026

The four-part structure

Nearly every good response has the same skeleton. Once your team internalizes it, drafting takes two minutes.

1. Name the specific problem. Not "we're sorry you had a poor experience." Say "you waited three days for a callback on a no-heat call." Specificity is the entire signal that a human read the review.
2. Take the position, briefly. Either it was a miss and you own it, or there's context and you state it once, calmly. One sentence. Everyone over-writes this part.
3. State the correction. What changed, or what you're offering. Vague regret with no action reads as boilerplate.
4. Move it offline with a name and a channel. "Call Dana at 555-0134" beats "please contact our customer service team."

Then stop. Long responses lose. The audience is never the reviewer — it's the next hundred people reading the profile while deciding whether to call your client, and they're scanning for one thing: does this business handle problems like an adult?

Speed is the metric your client actually feels

Response quality varies within a narrow band — most competent responses land in the same place with the same reader. Response time varies by an order of magnitude, from two hours to never.

Most people don't realize how much of the damage happens in the waiting rather than in the review itself. A one-star review sitting unanswered for three weeks costs something every day. The same review with a calm reply underneath it, posted the next morning, converts a chunk of readers toward the business, because they now have evidence of how it behaves when things go sideways.

Build for latency, not polish. Set an internal SLA of 48 hours on anything three stars or below and measure against it monthly. That number does more for a client's reputation than rewriting your templates ever will.

The bottleneck is the approval loop, not the writing

This is the nuance separating agencies who do reputation work well from agencies who technically offer it.

A review response is a public statement made in a client's voice about a customer's bad day, so someone at the client has to be comfortable with it. Route every draft through a busy plumbing company owner who checks email at 9pm and your 48-hour SLA is gone. That's why rewriting your response templates never fixes a slow program — the templates were never the constraint.

You don't need a dedicated reputation manager to hold a 48-hour SLA. Three fixes, in order of how well they hold up:

• Pre-approve the framework, not each response. Get the owner to sign off once on the structure, the offline contact name, and the tone. Then you only escalate exceptions.
• Define the exception list up front. Anything alleging property damage, injury, a billing dispute, or discrimination goes to the owner. Everything else ships under the pre-approved framework.
• Keep drafting off the critical path. The draft should already exist when a human opens the notification. If your process starts with a blank box, you've built latency into step one.

That last point is where tooling earns its keep. DFY Hub's Review Monitoring aggregates reviews across Google, Yelp, and Trustpilot and drafts AI responses, and the platform ships a Confirmation Required Flow that puts human approval in front of high-stakes agent actions before execution. The drafting is automated; the decision to publish under a client's name isn't, and shouldn't be.

I'd push back on any vendor pitching unattended review responses. The failure mode isn't a clumsy sentence — it's a system confidently apologizing for something that didn't happen, in your client's name, on the profile their phone calls come from. Keep the human step; it costs about a minute.

Three templates to adapt

A real service failure (1-2 stars, valid complaint)

Ben — you're right, and I'm sorry. We quoted a Tuesday window, didn't get a tech to you until Thursday, and nobody called to tell you it had slipped. We've changed how dispatch handles schedule changes so customers get a call the same day it happens. I'd like to make the delay right — Dana at 555-0134 can take care of it directly.

A pricing complaint (the work was fine, the invoice stung)

Thanks for the honest feedback, Maria. Our diagnostic fee is $89 and it's applied to the repair if you move forward — from the ticket, we didn't explain that clearly before the visit, and that's on us. I've asked the office to walk through pricing on the phone before we dispatch. If you want to go through the invoice line by line, Dana at 555-0134 will make time for it.

A review that isn't from a customer (mistaken identity or spam)

We take every review seriously, so we searched our records for this name, address, and phone number and couldn't find a service visit. It's possible this was meant for another company. Either way we'd like to sort it out — please call Dana at 555-0134. We've also flagged the review with the platform while we look into it.

Note the last one: you report it and you respond. A reported review can sit in limbo for weeks, and unanswered accusations read as admissions.

When to report instead of respond

Some reviews violate platform content policies rather than describing a bad experience — off-topic rants, competitor postings, personal attacks, content about a different business. Report those through the platform's own process.

Two things to tell clients so they don't build false hope: reporting is slow, and "this review is unfair" is not a policy violation. Removal happens for policy breaches, not for being wrong or harsh. Check the current policy language before promising an outcome.

Don't run this only on Google

Your client's Google profile is where the phone calls come from, so it gets the attention. But an unanswered one-star Yelp review is visible to anyone searching the business name, and Trustpilot shows up for larger regional operators. Monitoring all three is table stakes in an agency scope of work — the alternative is a client forwarding you a screenshot of something that's been live for six weeks.

For the wider program, reputation management for service companies covers the full picture, and the intake side is in how to get more google reviews. That second one is the strategic piece: the cheapest defense against negative reviews isn't a better response, it's enough recent positive volume that one angry customer moves the average by a rounding error.

Showing the client it's handled

Response rate belongs on the monthly report next to rankings and traffic — it's the reputation metric an owner intuitively understands: what share of reviews actually got a reply. Average time-to-response isn't a built-in report metric yet, so if a client specifically asks for that number, track it yourself for now rather than promising it shows up on the automated PDF.

The Client Reporting Portal supports branded PDF reports with AI executive summaries, share links, and scheduled delivery, drawing on 15 data modules across 8 templates — response rate, review count, average rating, and platform breakdown are among them. Put it on a schedule and the client stops wondering whether anyone's watching the profile. Tier details are on the pricing page.

FAQ

How fast should you respond to a negative review?

Within 24 to 48 hours for anything three stars or below. The exact hour matters less than having a standing deadline you measure against, because the realistic failure mode isn't a slow response — it's a review nobody noticed for a month.

Should you ever respond publicly to a false or defamatory review?

Yes, once, calmly, while also reporting it through the platform. Silence reads as confirmation to the next reader, and removal is slow enough that you can't rely on it. Keep it factual and short, don't accuse the reviewer of lying, and never disclose account details to prove they weren't a customer.

Can AI write the responses for our client accounts?

It can draft them. DFY Hub's Review Monitoring includes AI response drafting, and the platform's Confirmation Required Flow requires human approval for high-stakes agent actions before they execute. Treat the output as a first pass a person edits and approves — a public apology issued in a client's voice isn't something to publish unread, however good the draft.

What should never appear in a public review response?

The customer's account details, address, or service history; an argument about who's right; anything defensive about a competitor; and any offer that reads as paying for the review to be changed. Offer to make it right offline — don't negotiate the star rating in public.