The lead is the product

Most marketplaces sell a thing or a booking. Some sell neither: what changes hands is a customer's stated intention to buy, and a supplier pays for the right to contact them.

That single difference reshapes the whole system. A lead cannot be inspected before purchase, cannot be returned in any meaningful sense, and is worth nothing five days after it was created. The platform is selling a perishable good whose quality the buyer can only assess after paying.

Moderation is the load-bearing part

Between the customer submitting an enquiry and a supplier receiving it sits a check: is this real, is it complete enough to act on, is it in a category anyone serves.

That check is squeezed from both directions. Hold a lead too long and the customer has already found someone else, so the supplier pays for a conversation that goes nowhere. Release it unchecked and suppliers pay for junk — a mistyped number, a test submission, someone who wanted a price and nothing else.

Both failures look the same to the supplier. They paid and got nothing. Enough of those and they stop buying, and a marketplace with no buyers on the supply side has no business.

So moderation cannot be an afterthought queue. It needs to be fast by design: automated where the signal is clear, human where it is not, and instrumented so that the time a lead spends waiting is a number somebody watches.

Exclusivity is the decision everything else depends on

Is a lead sold once, or to several suppliers at once?

Sell it once and it is worth more, suppliers trust it, and the platform earns less per enquiry. Sell it five times and revenue multiplies while the customer receives five calls about one job and each supplier's odds fall to one in five. Suppliers work this out quickly, and what they conclude sets the price they will tolerate.

Neither answer is wrong, but it has to be decided deliberately, disclosed, and enforced by the system rather than by policy. It also has to be modelled from the start: retrofitting exclusivity onto a schema that assumed unlimited distribution means rebuilding pricing, crediting and reporting at once.

The incentive that has to be designed against

The platform is paid per lead sold. Suppliers need leads that convert. Those two things point in opposite directions, and every mechanism in the product either mitigates that or makes it worse.

The honest mitigations are structural: a credit or refund process for leads that were genuinely bad, quality signals that affect what a lead costs, and reporting that lets a supplier see conversion rather than only spend. Each of them reduces short-term revenue, which is exactly why they get postponed, and postponing them is how a marketplace acquires a reputation it cannot shed.

Refunds require an adjudicator

Once suppliers can claim a refund, the platform is judging disputes about something unobservable — whether a customer was reachable, whether the enquiry was serious.

Without a record, this becomes a negotiation decided by persistence, which rewards the wrong suppliers. With a record — when the lead was released, whether contact details validated, how quickly the supplier acted, whether the customer engaged elsewhere on the platform — most claims answer themselves.

That evidence has to be captured as a by-product of normal operation, because it does not exist retrospectively. A refund policy is only as good as the data available when someone invokes it.

Fraud runs in both directions

Fake enquiries drain a competitor's credit balance or inflate apparent demand in a category. Suppliers claim good leads were bad. Neither is exotic; both appear as soon as there is money in the system.

Validating contact details before release removes most of the first, and rate limiting by source removes more. The second is handled by the same evidence trail that handles honest refund claims, which is the argument for building it before it is urgent.

The metric nobody has

The platform can see a lead released and a credit spent. What it usually cannot see is whether the supplier won the job, because that happens off-platform in a phone call.

Every marketplace like this eventually tries to close that loop, because without it the platform is optimising spend rather than outcome. It is worth deciding early how much of that loop to attempt, since the answer shapes what data you ask suppliers for from day one — and asking later, once habits are set, rarely works.