Ask every buyer once, and chase none of them

The review you should not chase is the follow-up: ask once, at delivery, in the same words on every job, and never nudge. The tempting alternative - skipping the ask on jobs that went awkwardly - is now a regulatory risk in the UK, because encouraging only satisfied buyers to review is treated as cherry-picking.

Why the nudge is the ask to drop

A buyer who has not replied to the delivery has finished with the job. A nudge turns a non-event into a prompt to have an opinion, and prompted opinions are the flattest reviews in the trade. It also lands unevenly: the buyers who ignore a first request are the ones least engaged with the work, so a second request samples exactly the people with least to say.

A missing review costs you one increment on a count. A specific negative review costs that increment plus text a buyer will read in full, because negative text is what gets read in full, sitting near the top of your listing for months. Chasing trades a small, certain nothing for a chance at the second outcome.

Why not just skip the awkward ones

The obvious refinement is to ask on the jobs that went well and stay quiet on the ones that did not. It is also the one the UK's Competition and Markets Authority names. Its April 2025 guidance on fake and misleading reviews (CMA208) lists "encouraging just those who are satisfied to leave reviews" as a form of cherry-picking, and warns against "arbitrarily stopping and starting review invitations". The rules differ by country and this is not legal advice, but a practice a regulator has named in writing is a poor foundation for a listing.

The consistent version is also the easier one. One sentence in every handover, as set out in the single ask, with no decision attached.

Deliver the awkward job normally

The jobs that end badly are recognisable: a scope push you held, a brief and a delivery that never quite met, a negotiated discount, a late delivery on your side, an exchange that got personal.

Deliver them exactly as you would any other: full quality, on time, the same handover message including the same review sentence. Do not deliver early to compensate, do not add unrequested work as an apology, and do not write a longer message. Each reads as anxiety, and anxiety invites a response.

If a fair negative review arrives, it is information about the job, and a short, factual reply where the platform allows one reads better to the next buyer than the review alone. Where money is the real grievance, one refunded fee often costs less than the alternative, because the fee is the small number in the equation.

Fixing the upstream problem

Most of these jobs were identifiable before they started: a thin brief, a haggle in the first message, a request that half-matches your listing. Two screening questions before you accept catch most of them, and the questions that save you an hour are the cheapest intervention in this trade.

If more than roughly one job in ten ends awkwardly, the problem is not the buyers. It is a tier, a listing line or a price that brings in people expecting something else, and no handling of reviews fixes that.

Most awkwardness of this kind is mutual misunderstanding rather than bad faith, and the commissioning-side account of how a job is meant to open fairly describes what a buyer thinks they are doing. When the material itself was the problem, the notes on what makes something assessable are the thing to point at. And do not read much into one poor review; small samples of subjective scores support far less than they appear to, which is the standing caution about aggregate figures.

Where a platform sends its own automatic review prompt regardless of what you write, it is worth knowing which of your buyers get one; Rate Cock documents the judge side of that.