Automated review requests send every customer a short, polite ask after the job, with a direct link, and at most one reminder. AI helps with timing, wording and replying to reviews quickly. What it must not do is filter who gets asked by whether they seem happy, reward people for reviews, or produce reviews nobody actually wrote, because platform rules and federal law prohibit all three.
Reviews matter to how customers choose a local business, and most owners know they should ask more often. The problem is consistency. Asking happens when someone remembers, which in practice means after the jobs that went best. Automation makes the ask routine. This guide covers how to set it up so it helps you and stays within the rules.
What FiberX's reviews agent does
As described on our AI automation page, the AI reviews and reputation agent sends review requests after every job, replies to feedback quickly, and alerts you before one bad review becomes a pattern. The phrase that matters is "after every job": the request goes to every customer, not a hand-picked few. Like the other seven agents, it can be the first one you start with or an addition later, it needs no new hardware or phone number, and your dedicated agent sets it up.
The rules, in plain terms
General information, not legal advice. Platform policies change, so read the current rules for each site you use.
- No review gating. Google's review policies prohibit discouraging negative reviews or selectively asking only happy customers. A system that first asks "How did we do?" and sends only the satisfied ones to Google is gating, however it is dressed up.
- No incentives for reviews on Google. Google prohibits offering money, discounts, gifts or other benefits in exchange for reviews.
- The FTC's rule on reviews. A federal rule in force since October 2024 bans fake reviews, including ones attributed to people who did not have the experience, buying positive or negative reviews, offering incentives tied to a review's sentiment, undisclosed reviews by owners or staff, and suppressing reviews through threats or false accusations. Penalties can be significant.
- Yelp is different. Yelp asks businesses not to solicit reviews at all. Leave Yelp out of your automated requests.
- Healthcare has extra limits. A medical or dental practice should not confirm in a public reply that the reviewer is a patient, or mention anything about their care. Reply generally and invite them to call.
- Texting the ask needs consent. If requests go by text, collect consent, identify your business and include opt-out instructions. Business texting from software also has to be registered with the carriers.
| Fine to do | Not fine |
|---|---|
| Ask every customer after the job, the same way | Ask only customers you expect to be happy |
| Include a way to reach the owner directly, as well as the review link, for everyone | Send unhappy customers to a private form instead of the review link |
| Send one polite reminder | Message repeatedly until someone reviews |
| Use AI to draft your replies, approved by a person | Use AI to write reviews, or "suggested review text" for customers to post |
| Reply to every review, good and bad | Offer a discount for a review, or for changing one |
Building the request
Timing
Ask when the work is done and fresh in the customer's mind: the same day for a service visit, after delivery for a product, after the first visit for an ongoing service. A request three weeks later feels like an afterthought, and it is.
Channel
Text tends to be read quickly, where the customer has agreed to texts. Email works for everyone else and suits businesses whose customers book by email anyway. Use one channel per request, not both at once.
Wording
Short, specific and without pressure. An example for an invented plumbing company:
Hi Dana, thanks for having Eastside Plumbing out today. If you have a minute, a review helps neighbours find us: [link]. If anything was not right, reply here or call me directly. Mike, owner. Reply STOP to opt out.
Every customer gets the same message, including the line about calling the owner. That line is the honest alternative to gating: unhappy customers can reach you directly, and they can still review you if they choose.
Follow-up
One reminder a few days later, then stop. Stop immediately for anyone who opts out or replies that they would rather not.
Replying to reviews
Replies are where AI saves the most time, and where a person should still have the last word.
- Positive reviews: a short, specific thank-you that mentions what the customer mentioned. AI can draft these, and many owners let them post after a quick look.
- Negative reviews: acknowledge the problem, do not argue in public, offer a direct way to sort it out, and leave out private details. AI drafts; a person approves before anything posts.
- Reviews you believe are fake: report them through the platform's own process. Do not threaten the reviewer.
Patterns are the real value
A single bad review is a bad day. The same complaint three times in a month is an operational problem: late arrivals, a confusing invoice, one location's phone that never gets answered. An alert that groups complaints by theme gives you something to fix, not just something to reply to. If the theme is calls going unanswered, see the real cost of a missed call.
A worked example
Hypothetical: the company, job counts and response rate are invented to show the arithmetic. Your results will differ.
A plumbing company completes about 120 jobs a month. Today, technicians ask a handful of customers in person, usually the friendliest ones, and the company gets 2 or 3 new reviews a month.
With automated requests, every completed job triggers a text or email within two hours, and a single reminder three days later. If 15% of customers left a review, that would be 18 a month. The owner reviews AI-drafted replies for about 3 minutes each, under an hour a month in total.
The bigger change is in what the reviews say. Because every customer is asked, the reviews reflect the whole month, including the jobs that went less well. In the second month, three reviews mention arriving outside the promised window. The alert flags it, the owner changes how arrival windows are scheduled, and the next month's reviews stop mentioning it. That is the useful kind of feedback loop, and it only works because nobody was filtered out.
Connecting it to the rest of the business
Review requests work best when triggered by something that really happened: a job marked complete, an appointment checked out, an order delivered. That means connecting the agent to your scheduling, job or point-of-sale system, rather than uploading lists by hand. It also means the request can use the customer's name and the right location's review link. If you are already using automated follow-up for new enquiries, the same consent records and opt-outs should apply to review requests; see AI lead follow-up.
Several locations or several technicians
A business with more than one location should send each customer to the review page for the location that served them, not to a single head office listing. Reviews then help the branch customers actually visit, and complaints point at the place where the problem happened. If customers know their technician or stylist by name, the request can mention them, which tends to read as more personal. Keep one rule across every location, though: the same message, to every customer, with the same link.
What to track
| Measure | Why it matters |
|---|---|
| Requests sent against jobs completed | Proves every customer is being asked, which is also your evidence that nothing is gated |
| Reviews received per month | Shows whether the ask is landing |
| Time to reply | Customers read replies, especially to complaints |
| Opt-outs | A rising number suggests the wording, timing or frequency is wrong |
| Repeated complaint themes | The operational problems worth fixing first |
The first row is the one to check monthly. If requests sent falls well below jobs completed, something in the trigger is broken, or someone has started choosing who gets asked.
Questions to ask your provider
- Does the system send requests to every customer, and what exclusions does it allow? Do those exclusions comply with each platform's rules?
- Which review sites does it send people to, and can we leave out sites that ask businesses not to solicit reviews?
- How are text consent, opt-outs and texting registration handled?
- Which replies post automatically, and which wait for our approval?
- How does it flag repeated complaints, and who receives the alerts?
- For healthcare clients, how are replies kept free of patient information?
- What triggers a request: a manual list, or an event in our scheduling or job system?
Getting started
If you want to see what the reviews agent would look like for your business, the consultation is free with no obligation, and Phil Morales sets it up as your one point of contact. Many clients start with the phone, using our AI receptionist, and add reviews once they trust it. Call 478-758-8091, where our own AI receptionist answers at any hour, text (347) 870-0965, or book a time. You usually hear back the same day.
