Asking for a review is the most awkward thirty seconds of the job

An electrician showed me his Google profile last year. Fourteen years in business, work he was genuinely proud of, a customer list with people on it he had been back to four or five times. Eleven reviews.

Meanwhile a company two towns over, who he was fairly sure were cutting corners, had two hundred and something and was appearing above him in the map results for every search that mattered. That is the part that gets people. Not vanity. The fact that a stranger comparing two names on a phone screen has no way to tell the difference between a good business with eleven reviews and a mediocre one with two hundred, and will pick the two hundred every time.

He was not bad at his job. He was bad at one specific and genuinely uncomfortable moment, and so is nearly everybody.

Why the driveway ask fails

Picture the actual moment. You have finished. You are carrying something heavy. The customer is standing there with their card out or their phone in their hand, and you are both doing the small ritual of ending a transaction. And somewhere in that thirty seconds you are supposed to say, cheerfully, would you mind writing something nice about me on the internet.

Three things go wrong. You forget, because you are thinking about the next job and the traffic. Or you remember and it feels like asking for a favour, so you say it apologetically, in a way that makes it easy to say yes to and easy to never do. Or you say it perfectly, they mean it, and then they get in from work at half six and it is gone from their head forever.

The intention to leave a review has a half life of about an hour. Sometimes less. Almost nobody who does not do it was being dishonest with you.

Timing beats wording, and it is not close

People obsess over the words in the request. The words matter a bit. The timing matters enormously.

The wrong time is when you remember, which is usually when you are doing your invoices on a Sunday evening, which is also roughly the worst hour of the week to land in somebody's messages. The right time is a couple of hours after the job is done, when the relief is still there and the thing you fixed is still noticeably fixed, but they are not standing next to you feeling put on the spot.

Beyond that, timing is about their day rather than yours. A request that arrives while somebody is on the school run gets swiped away. The same request at eight in the evening gets read. So it goes out when it is likely to be read, on completion, without anybody having to think about it. That single change, moving the ask off your memory and onto the job being marked complete, does most of the work in this whole system.

One ask, one nudge, then leave people alone

Our whole sequence is two messages. The request, and then one follow up a few days later if nothing happened. That is it. After that the customer has answered by not answering, and continuing is just making yourself unwelcome.

The follow up is short and gives them an easy exit. Something like: no problem at all if you would rather not, here is the link once more in case it got buried. That sentence increases replies rather than reducing them, which is one of those things that sounds like a nicety and is actually just how people work.

We get asked for a third and fourth reminder occasionally, and we say no. Not on principle exactly, but because a business that nags is remembered as a business that nags, and you are trading a small number of extra reviews against that.

The unhappy ones, and the thing you cannot do

Here is where a lot of tools quietly break the rules, and where you should be careful about what you agree to.

The tempting design is obvious. Ask the customer how it went first. If they say great, show them the Google link. If they say not great, send them to a private feedback form instead and never mention the review. Everybody selling this software has built that flow at some point, because it works and it is easy.

It is also review gating, and Google's policy on it is not ambiguous. You are not allowed to selectively solicit reviews only from customers you expect to be positive. The consequences are not usually a dramatic ban, they are worse than that in a way: reviews get filtered, ranking quietly suffers, and you have no idea why because nothing announced itself.

So the request goes to every customer, the same way, regardless of how we think the job went. Nobody is filtered out.

What you can do, and should, is make sure there is an obvious private route to a human at the same time. Every customer gets both: a review link and a plain sentence saying if something was not right, reply to this and it comes straight to us. Not a fork in the road based on a rating. Two doors, offered to everybody, and one of them is a person who can fix it.

In practice that recovers a decent share of the reviews that would otherwise have been one stars, not by hiding them but by fixing the thing first. Someone whose real complaint is that you left the driveway a mess mostly wants the driveway sorted. Give them somewhere to say that at nine in the evening and a lot of them will say it to you rather than to the internet.

What ninety days actually looks like

I want to be careful here, because this is the section where most agencies produce a number and most numbers are decorative.

What I can tell you honestly is the shape rather than the multiple. A business that was asking sometimes, verbally, and getting one review a month, tends to end up getting reviews from a meaningful fraction of completed jobs, consistently, because the ask now happens every time instead of when someone remembers. Whether that is four a month or forty depends entirely on how many jobs you complete, which is your number and not ours.

The second order effect is the one people notice more. Recent reviews are worth more than old ones to the people reading them, and a profile with something from last week reads as a business that is alive. Eleven reviews where the newest is from 2023 reads as a business that might have closed.

Where this does not help

• If the work is not good, this makes the problem public faster. That is not a bug, but it is worth saying plainly, because a few people arrive hoping for reputation management and this is not that.
• Trades with genuinely private customers. Some commercial and legal work has clients who will not review anything publicly, ever, and no amount of good timing changes it.
• Anything with a long gap between the work and the result. If the customer cannot tell whether you did a good job for six months, asking on completion is asking them to guess.
• Businesses with very few, very large jobs. Six jobs a year means six chances, and at that volume a human being should be asking personally rather than a system doing it for them.

The other half of this is what happens to the reviews once they arrive, which is its own piece. Getting them is the easy part. Answering all of them, including the bad one, within the hour, forever, is the part nobody keeps up.