Review policy

The question every review answers

Did they reimburse me, in full, when they said they would — and were the terms what they promised?

That is what a TrustDrive review is for. It is not a general opinion about a mediator and it is not a star rating for a shop. Each review records what actually happened with one campaign: whether you were paid in full, paid late, paid short, or not paid; how long the money was out of your pocket; how reachable they were; and whether the terms held.

Who can write one

You cannot review a mediator you have not worked with. This is the strongest control the product has, and it is the reason the reviews here are worth reading. To write a review, all of the following must be true:

• You declared the campaign on TrustDrive — the mediator, the platform, the deliverable, what you ordered, what it cost, and the dates. That declaration is the record everything else hangs from.
• The mediator has had their turn on it. After you declare a campaign the mediator has 14 days to confirm it or contest it. You can review once they have confirmed, or once that window has passed and the campaign is marked as unconfirmed. Their silence does not veto your review; it just labels the campaign as unconfirmed wherever it is shown.
• The money question has settled — you were paid, or short-paid, or you have concluded you will not be paid. If you are still waiting, the mechanism for speaking is a non-payment report, not a review.
• You are inside the review window, which is 180 days from the point the campaign becomes reviewable. Reviews of very old work mislead readers, and an unbounded window leaves a loaded weapon pointed at a mediator somebody stopped working with years ago.
• Your identity is verified to TrustDrive — a phone number at minimum — before you submit, not before your review is published.
• You have not already reviewed this campaign. One review per person per campaign.

If you have no declared campaign, the review form is replaced by the declaration flow rather than by an error. That is the intended route in, not a workaround.

Your name is not on it

Reviews publish under your display handle, not your legal name. In this market you are often writing about someone who still holds your money, and the realistic retaliation is not a lawsuit — it is being quietly dropped from every WhatsApp group in the region. A product that required your name on the report would only ever collect reports from people with nothing left to lose.

The honest limits, which we state rather than gloss:

• This is not anonymity from the mediator. In a twelve-person campaign group, a review naming the platform, the deliverable and the week identifies its author immediately. What the handle protects is the public record, other mediators, and search engines. Do not assume more.
• TrustDrive knows who you are. Your identity is verified before you can submit and is visible to staff. Pseudonymous is not unaccountable.
• You may choose to show your real name.
• TrustDrive will only reveal your identity to a mediator where a dispute genuinely cannot be decided otherwise — and you are told first, and offered the choice of withdrawing the review instead. That trade is yours to make.

When a review publishes, and when a person reads it first

Most reviews publish when you submit them. Human review of every review does not scale, and a queue nobody can keep up with becomes a rubber stamp, which is worse than an honest rule. So the rule runs the other way: a review publishes by default, and specific signals send it to a person first.

A review is held for a human when any of these is true:

1. It says you were not paid, or paid short. These are the statements most damaging if wrong and most important if right. A "not paid" submission leaves the review path entirely and becomes a non-payment report, which never publishes automatically.
2. It is the first review on a listing, from an account younger than 7 days. That is the shape of a manufactured attack and of a manufactured endorsement equally, so the rule does not care which direction it points.
3. It arrives in a cluster — an unusual burst of reviews, declarations or reports against one mediator, measured against that mediator's own normal rate rather than a global number.
4. There is an open report on the mediator, or they are disputing the listing itself. While somebody is contesting the money or the page, nothing new publishes against them unread.
5. It carries an allegation tag such as fraud or abuse.
6. It scores above the anomaly threshold, or you are inside a cooldown.

A held review is not a rejected one. You can see it, marked as under review, the whole time.

Editing, and withdrawing

• You can edit your own review for 24 hours after it publishes. An edit sends it back through moderation and snapshots the previous text.
• After that window the review is fixed and you can add a dated note underneath. The note is appended; it never replaces what you wrote. Silently rewriting a published statement about a named person, after they have already answered it, is the shape of both a laundered accusation and a bought retraction.
• You can withdraw a review at any time. It is unpublished, and the fact of the withdrawal and its date are recorded — a mediator with eleven reports and nine withdrawals is a pattern that no single withdrawal shows.
• Review length is between 40 and 2000 characters, and you may attach up to 10 files. Attachments are private: staff only, never published.

Why a review may be rejected

Every rejection names its reason and you are told what it was. The reasons are a fixed list, not a free-text judgement: no qualifying campaign; you were not the party to it; the conduct described belongs to a different mediator; off-topic; abusive content; it discloses somebody else's personal information; it makes an unsupported allegation of a crime; it is part of a detected coordinated pattern; you have a conflict of interest; it duplicates a review you already wrote; it was filed during a cooldown; or you withdrew it.

A rejected review is kept, never hard-deleted, and you can see it with its reason. If you fix the defect or successfully appeal, it goes back into the queue.

The mediator's right of reply

A mediator can reply publicly to any review about them, once. The reply is displayed with the review, permanently, and cannot be edited afterwards — a right of reply that can be quietly revised is a way to rewrite history after the fact. It can be redacted only for abuse, personal information, defamation or a policy breach, and the redaction is marked.

Built for claimed listings. A mediator who has claimed their listing and verified their own account's phone number can post the reply described above from their profile. A mediator whose listing nobody has claimed yet cannot reply this way today — not because the reply mechanism works any differently for them, but because TrustDrive has no lawful way to tell an unclaimed listing's owner a review exists in the first place (see Disputes, below). Claiming the listing is what opens the door, the same way it does for a dispute.

A mediator does not need to have claimed their listing to reply to a report about them. Requiring somebody to make an account before they can answer an accusation would guarantee they never answer.

Disputes

A mediator who thinks a review is wrong can dispute it. The grounds are a fixed list: it is fabricated; the author was never in their network; it is factually incorrect; it is defamatory; the author has a conflict of interest; it is retaliatory; it contains personal information.

Disputing a review requires a claimed listing, and replying to a report does not. That difference is deliberate. A report is answered to stop a clock, so we accept an answer from whoever holds the number we wrote to. A dispute is a formal record that a court may one day read, so it has to name who filed it — and today the only way to establish that is to claim the listing, which is a one-time code to a number already on the profile. Until a listing is claimed, the mediator it names cannot contest a review about them, and we say so on the listing itself rather than letting the silence read as agreement.

Filing a dispute does not take the review down. The review stays up with a visible "under dispute" marker while it is decided. If filing could unpublish a review, filing would be the way to suppress reviews, and every mediator would learn that within a week. A review is only taken out of view while a dispute runs if a moderator affirmatively decides the text may be defamatory, may expose somebody's personal information, or puts a named person at risk — and that decision is recorded.

A dispute can end six ways, and they are not one outcome with six labels:

• the review stands as written;
• the review is unpublished, and the dispute record stays visible;
• a specific claim is struck and the rest of the review stands;
• partly upheld;
• no action, insufficient evidence;
• not applicable.

Blocking disputes — those that would stop a review publishing — are targeted at 3 days. Disputes against an already-published review are targeted at 7 days. A mediator's dispute rate is itself a signal we watch.

Averages, and when we refuse to show one

Individual reviews show from the first one. Averages do not. Below a floor of 5 settled campaigns from at least 3 different influencers, TrustDrive shows the reviews and shows no aggregate at all — not a zero, not a provisional number, nothing.

A zero would be a claim, and a false one: it would rank a brand-new honest listing below a mediator with a genuinely bad record and it would state a measurement nobody took. The second half of that floor — different influencers, not just different campaigns — is what stops one person with a few logins manufacturing a record for a mediator they never worked with.

The reimbursement rate itself, and how it is calculated, is on the trust page. No number on this site is ever synthesised, and if there is no data the page says there is no data.

Manipulation, in both directions

Reviews can be faked to attack a mediator as easily as to promote one, and this product treats both as the same problem:

• identity is verified before you can submit anything;
• every review is anchored to a declared campaign with a value and dates;
• bursts against a single mediator are detected against that mediator's own baseline, in either direction;
• there are rate limits on how much anyone can file, and against whom;
• nothing that has not been adjudicated moves a mediator's position in the directory. An unanswered report, an aging count, a pile of one-star reviews: none of them reorder anything. Only a finding does.

Pressure runs the other way too. Conditioning reimbursement, or continued access to campaigns, on what you write here — or on taking it down — is a prohibited practice and grounds for enforcement. If that has happened to you, tell us.

Not decided yet. Whether a mediator who pays late but does eventually pay should be recorded as a resolved non-payment report or simply as a "paid late" review with a long delay attached. Today it is the second: if you were merely paid slowly you can rate it, and if you have been left waiting you can file. The owner has to confirm that before launch, because it changes how often flags appear.

Related: the non-payment report policy, the listing policy, and how the numbers are worked out. Changes to this page are recorded in the policy changelog.