The asymmetry that creates the problem.
A customer who had a good experience got what they paid for and has no particular reason to tell the internet. A customer who had a bad one is motivated. Left alone, that asymmetry produces a public rating worse than your actual service, which is why well-run businesses are frequently surprised by their own profile.
The fix is not to suppress the negative. It is to stop leaving the positive to chance.
Timing, and why it should not be a person’s job.
Ask while they still remember you — soon after the work is complete, not at the end of the month. And do not rely on a technician or an office manager to remember, because they will remember when they are not busy, which is precisely when you have fewest jobs.
Trigger it from the system that already knows the job finished. That single change usually does more than everything else on this list combined.
Give unhappy customers somewhere to go first.
A short check-in before the public ask — "how did we do?" — routes the unhappy ones to you instead of to a profile. That is not filtering, and the distinction matters: everyone still gets asked for a public review. You are simply finding out about problems in time to fix them.
Filtering who gets asked based on predicted sentiment is a policy violation and platforms detect it. Asking everyone, and giving everyone a route to complain directly, is not.
Recency beats volume, past a point.
Nobody reads your average. They read the three most recent reviews and how you replied to the worst one. Four hundred reviews with the newest from two years ago reads as a business that used to be good.
This is also why a one-off review push is a poor investment: it produces a cluster, then silence. A steady trickle from completed work is worth more and looks more like what it is.
Answer the bad ones, quickly and without arguing.
The audience for your reply is never the reviewer. It is the next customer, who is reading to find out how you behave when something goes wrong.
A calm, factual, non-defensive reply to an unfair review frequently does more good than the review did harm. An argument does the opposite, permanently, in public.
And if the reviews are accurate, more asking will simply publish more of the same, faster. That is an operations problem and no reputation tool fixes it.