GBP review requests: ask every customer after a job or only the happy ones?

Local service shops split on this all the time. Some send the review link after every finished job, others only ask when the tech thinks the customer was happy. Take a plumbing company doing 40 jobs a month. If they only ask the happy ones, the star average looks great, but the profile grows slowly and the local pack has fewer fresh reviews to work with. Asking everyone gives you a profile that looks like the real business, and the complaints that do come in are usually fixable. The one thing I'd treat as a hard line is filtering by sentiment, because Google's policy prohibits selectively soliciting positive reviews. So what's your rule: does the office send the link to everyone, or does the tech decide who gets it?

Agreed on the hard line, and I'd go further. The tech's gut feeling is the wrong filter even before policy comes into it, because techs are bad at predicting who'll leave a bad review. A lot depends on whether the office can send the link automatically after job close. If it can, asking everyone costs nothing and nobody has to decide.

I wouldn't call it free. If the link goes out automatically the second a job closes, an unhappy customer gets a review prompt before anyone has had a chance to fix the problem. Plumbing jobs often turn up a leak or a callback two days later, and a one-star review posted in the first hour is a lot harder to recover from.

I'd still send to everyone, just with a delay of a day or two so the office hears about trouble first. Keep that delay the same for every customer, though. Once you gate it on how the job went, you're back in the filtering problem.

Agreed on the delay, but there's a back door to watch for. If the office pulls a job from the send queue because a callback is open, you're still gating on how the job went, you're just not calling it filtering. The queue should stay dumb, and the callback gets resolved on its own timeline. That also makes it measurable: compare sends against closed jobs each week, and any gap shows up as a number instead of a judgment call. Where do callbacks get logged today, in the job software or in someone's head?

Honestly, it comes down to how long a typical callback takes to close. If most fixes happen the same day, a fixed two-day delay covers them and the queue can stay dumb, like you said. Plumbing has plenty of jobs where the callback needs a part or a second visit that takes a week, though. For those, the fixed delay runs out before the problem does, and the pressure to hold individual jobs comes right back. I'd widen the delay for the whole queue in that case instead of letting the office pull jobs by hand, because that hand-pulling is exactly the gating you're trying to avoid.