Review request automation, done properly and legitimately

Nearly every business intends to ask for reviews and nearly none do it consistently. Automating the ask at the right moment is unglamorous and it works.

The reason it does not happen manually is the same reason quote follow-up does not happen manually. The right moment to ask is the day the job finished, and on the day the job finished the business owner is already at the next job. By the time they think about it, the moment has passed and asking a fortnight later feels odd.

So the ask gets skipped, and a business with hundreds of satisfied customers has eleven reviews, four of them from 2019, and one from a man who was annoyed about parking.

The rule that governs everything else

Before any of the practical detail, there is one rule, and it is not negotiable.

You ask every customer. You do not filter by expected sentiment.

There is a widely-sold pattern where a system first asks the customer to rate their experience privately, then sends the happy ones to a public review platform and routes the unhappy ones into a private feedback form instead. It is often marketed with words like "reputation protection".

Do not build this. Do not sell it. It is review gating, and it breaches the policies of the major review platforms, including Google's. Businesses caught doing it can have reviews removed and face further consequences. It is also, plainly, a way of manufacturing a rating that does not reflect reality, which is the sort of thing that reads badly in front of a regulator and worse in front of customers.

The line is straightforward. A system that sends the same request to every customer who received the service is legitimate. A system that decides who gets asked based on how they are likely to answer is not.

Specifically, the build must not:

• Screen for sentiment before deciding whether to send a review link.
• Show a review link to some customers and a private feedback form to others based on their response.
• Suppress, delay or divert negative feedback to keep it off a public profile.
• Offer discounts, entries, credits or any other incentive in exchange for a review, and certainly not for a positive one.
• Draft the review text for the customer or supply a wording to copy.

Collecting private feedback is fine and often valuable. It just has to be a separate thing, offered to everybody, not a diversion channel for the people you would rather were not heard.

What triggers a request

The trigger is completed, delivered service. Not a booking. Not a deposit. The customer has to have actually received the thing they are being asked to review. The booking stage has its own reminder sequence, and it ends where this one begins.

What that looks like in practice varies:

• A garage: the car has been collected.
• A salon: the appointment has been completed and paid.
• A plumber: the job is marked complete, or the invoice has been settled.
• A driving instructor: a test has been passed, or a block of lessons has finished.
• An estate agent: completion, not offer accepted.

Exclusions matter as much as the trigger. Do not send to a customer whose job is in dispute, whose invoice is contested, who has an open complaint, or who has already been asked recently. A customer who has left a review should not be asked again for a long time, if ever.

That last exclusion needs some judgement, because a regular customer - a monthly cleaner, a fortnightly driving lesson - cannot be asked after every visit. Once, after an early positive experience, then leave it alone. The same restraint applies when getting back in touch with customers who have gone quiet.

Timing

Same day or next day, while the experience is still current and specific. A customer asked on the day can remember the name of the person who did the work and what they said. A customer asked three weeks later writes "good service, would recommend", which is worth much less to anyone reading it.

Some sensible variations:

• Immediate service, visible result - a haircut, a valet, a repair - ask within a few hours.
• Work that needs proving - a boiler installation, a roof repair before wet weather - wait a few days. Asking for a review of a boiler an hour after it was fitted is asking about the installer's manner, not the work.
• Long transactions - a house sale, a large project - ask at the end, when the relief is fresh.

Avoid sending in the evening or at weekends unless that fits the business. A review request at eight on a Sunday night is a small intrusion.

What the message says

Short, specific, personal, one link.

Hi Priya, thanks for having us out today - hope the shower's behaving itself now. If you've got a minute, a quick review would really help us out: [link]. No worries if not. Dan, Fairfield Plumbing. Reply STOP to opt out.

What makes that work:

• It references the actual job. The customer can tell it was sent because of them.
• It is honest about why. "It really helps us out" is true and people respond to it. Small businesses get a lot of goodwill they never draw on.
• It has one link. Not three platforms. Pick the one that matters most for the business - usually Google - and use it. A choice of platforms reduces completions.
• It gives an easy out. "No worries if not" costs nothing and prevents the request feeling like an obligation.
• It is from a person. Signed by the individual, not the entity.

What must not be in there: any suggestion of what to write, any mention of a star rating you would like, any incentive, and any framing that implies the customer owes you something.

The link

Take the customer straight to the review form. Every extra tap loses people, and the population you lose to friction is not random - it is skewed towards the less motivated, which for reviews tends to mean the merely satisfied rather than the delighted or the annoyed. Use the direct link the platform provides.

Reminders

One reminder. That is the whole policy.

Send it three or four days after the first request, only to people who have not left a review and have not opted out. Keep it shorter than the original and lead with the acknowledgement that you are asking twice.

Hi Priya, sorry to nudge - if you get a spare minute the review link's here: [link]. Last time I'll ask. Thanks, Dan.

"Last time I'll ask" is a commitment. Keep it. A third request is where a review sequence stops being an ask and starts being an irritation, and an irritated customer is capable of leaving exactly the review you did not want. That failure mode - automating your way into a bad review - is real and entirely self-inflicted.

Compliance and cost

A review request is a marketing communication in most sensible readings, so treat it accordingly. UK electronic marketing rules apply. Every message identifies the business, carries a working opt-out, and the opt-out is honoured immediately and permanently across all sequences, not just this one. Only send to customers who actually bought or received the service. Never buy a list, never message people who found you but did not use you.

Email and SMS both work. SMS gets opened more, and it costs per message. Two messages per completed job, across a business doing a decent volume of work, adds up. Some businesses use email for the first ask and SMS only for the reminder, which is a reasonable compromise. Whichever way, price it in.

Keep the platform rules in mind too. They change, they differ between platforms, and the person building the system is responsible for knowing the current position. The constant is that filtering by sentiment and paying for reviews are prohibited more or less everywhere.

Handling the negative ones

If you ask everybody, you will occasionally get a negative review you would not have got otherwise. This is the honest trade-off and it is worth taking, for two reasons.

First, a profile of nothing but five-star reviews reads as suspicious to anyone who has spent time online. A handful of mixed reviews among many good ones is more persuasive than uniform perfection.

Second, a negative review answered well is a public demonstration of how the business handles problems, which is exactly what a nervous customer is trying to work out before they let a stranger into their house.

So build the response side too. Notify the owner when a review arrives, especially a poor one. Draft nothing automatically - a generated reply to an unhappy customer reads as contempt. But make sure the business knows within the hour, because a same-day reply is worth far more than one a fortnight later. It is one of the clearest cases for handing a situation straight to a person.

What you do not do is try to suppress it, bury it under a burst of solicited positives, or channel that customer away from the platform in the first place.

How it fails

It asks the wrong people. Customers with open complaints, disputed invoices, or jobs that went badly. Exclusion rules are not optional.

It asks too often. Regulars getting a request after every visit will disengage, and some will say so publicly.

It is generic. No name, no job reference, no person signing it. It reads as a mass send because it is one.

Nobody watches what comes in. The requests go out, reviews arrive, none get a response. Half the value of a review profile is in the replies.

Someone builds a gate. Whether from ignorance or because a client asked, this is the one that carries real consequences. Refuse it, explain why, and offer the legitimate version instead. Being clear about what you will and will not build for a client is easier before the work starts than after.

Done properly, review automation is one of the simplest pieces to build and one of the most visible in its effect - a business whose profile reflects the actual quality of its work rather than the subset of customers motivated enough to review unprompted. The automations overview shows where it sits alongside the rest, what's included covers the components, and the FAQ deals with the questions this one tends to raise.