Google Reviews on Autopilot

Filtering out the unhappy customers before you ask is called review gating, and it is against Google's policy.

Most businesses handle Google reviews in one of two ways. They never ask, and end up with a handful of reviews accumulated by accident, skewed towards people who felt strongly enough to act unprompted. Or they ask badly, in a way that annoys customers or quietly breaches Google's policy.

Both are fixable, and the fix is mostly about timing and consistency rather than persuasion.

The tactic to avoid first

Review gating is the practice of surveying customers before you ask, and only sending the Google review link to the ones who answered positively. Everyone else gets routed to a private feedback form. Several review platforms have marketed this as a feature.

Google's prohibited-content policy for reviews forbids it. That alone is reason enough not to do it, but there is a second reason that matters more in the long run: it destroys the value of the thing you are building. A 4.9 average assembled from pre-screened customers does not mean what a reader believes it means. The whole point of a public rating is that it samples everybody. Filter the sample and you have manufactured a number that looks like evidence and is not.

The same reasoning rules out the other common shortcuts. Do not offer a discount, a prize draw entry or anything else in exchange for a review: incentivised reviews are prohibited regardless of whether you asked for a positive one. Do not write reviews for customers, even if they say they are happy for you to. Do not ask staff or family to review the business.

What remains is permitted and, done well, is enough: ask honestly, ask everyone, ask once.

Timing does most of the work

The single largest factor in whether someone leaves a review is when you ask, not how you phrase it.

The window that works is immediately after the work is finished and the invoice has gone out. The experience is fresh, the outcome is known, and the relationship is at its most positive point. Ask two weeks later and the specifics have faded; the customer no longer remembers what was good about it, so they either write something generic or do not bother.

This is precisely why invoicing is a good trigger. It is the moment the work is complete and, usefully, it is already an automated event in most businesses. You do not have to remember to ask, because something you already do can do the asking.

That is the design behind Kudos: emailing a customer their invoice through Xero triggers one friendly review request on your behalf. No new step for anybody, and the timing is right by construction rather than by discipline.

Ask once, and ask everyone

Two rules govern the asking, and both are constraints rather than features.

Everyone gets the same link. No branching on how happy you think the customer is, no pre-survey, no filter. Every customer receives an identical request pointing at the same place. This is the compliance requirement, and it is also the thing that makes the resulting rating mean something.

One ask per customer per year. This is a self-imposed cap and it is worth explaining. A second request converts poorly, and it converts poorly while costing goodwill, which is the worst combination available. There is also a reputational effect that is easy to miss: a business that repeatedly chases reviews starts to look like a business managing its reputation rather than earning it. One well-timed ask captures most of the available benefit.

There is a third rule that belongs with them: the automation must never be able to post a review itself. Not as a safety feature but as an architectural property. The tool sends a link. Only the customer can write anything.

What to do about bad reviews

You will get some. Assume it at the design stage, because a system built on the assumption that everyone is happy will produce panic the first time someone is not.

Reply publicly, promptly, and without arguing. Acknowledge the specific thing that went wrong, say what you have done about it, and offer to continue the conversation directly. Do not relitigate the facts in public, even when you are right, because the audience is not the reviewer.

The reply is read by far more prospective customers than the review is, and it demonstrates something no five-star average can: that you handle problems well. Most people assume something will eventually go wrong; what they are trying to work out is what you do then.

It is also worth saying that a profile of nothing but perfect scores reads as implausible to most readers, and there is a reasonable body of consumer research suggesting ratings slightly below perfect are trusted more than perfect ones. A few honest three-star reviews with good replies do not damage you. They make the rest credible.

The whole system, briefly

Put together, a compliant automated review programme is not complicated:

The trigger is an event you already have: invoice sent, job closed, matter completed.
The request goes to every customer, with the same link, with no pre-screening of any kind.
It goes once, capped, with no chase sequence.
The message is short, names the work that was done, and makes clear an honest review is what is wanted.
Bad reviews get a public reply within a couple of days.
The tool can send and it can never post.

None of that requires persuasion technique, and all of it is dull to implement, which is a good sign. The businesses that end up with strong review profiles are almost never the ones with the cleverest asking strategy. They are the ones that ask consistently, at the right moment, and then stop.

Key takeaways

Never filter who gets askedReview gating breaches Google policy. Everyone gets the same link, regardless of how happy you think they are.
Timing beats persuasionAsk right after the work is done and the invoice is sent, when the experience is fresh.
Ask once, and mean itOne request per customer per year. Repeat chasing is what turns a fair ask into a nuisance.

Frequently asked

What exactly is review gating?

Surveying customers first and only sending the review link to the ones who responded positively, while routing unhappy customers to a private feedback form instead. Google's prohibited-content policy for reviews forbids this practice, and it also invalidates the signal, because a rating built only from pre-screened customers does not mean what a reader assumes it means.

Can we offer a discount for leaving a review?

No. Incentivised reviews are prohibited under Google's policy, whether the incentive is money, a discount, or entry to a prize draw. This holds even if you do not condition it on the review being positive.

Is it acceptable to ask at all?

Yes. Asking customers to leave an honest review is explicitly fine. What is not fine is choosing which customers get asked, paying for reviews, or writing them yourself.

What if we get a bad review?

Reply to it, publicly, without arguing. A considered reply to a critical review is read by far more prospective customers than the review itself, and it demonstrates something a five-star average cannot. A profile of nothing but perfect scores also reads as implausible to most people.

Why cap it at one request per customer per year?

Because the second ask converts poorly and costs goodwill, and because a business that repeatedly chases reviews starts to look like one buying its reputation. One well-timed ask is close to the whole benefit.