Cost per hire

Upwork's stats and trends page counts hires as proposals that "led to a contract", and the Connects page lists what was spent, but Upwork does not divide one by the other. An agency that wants to know whether Upwork pays has to.

There are two honest versions.

• Cost per win divides Connects spent on proposals by contracts won from them. It answers whether bidding is priced right.
• Acquisition cost divides all spend in the period, application fees, boosts, profile and message ads and the tool's own subscription, by every contract won from any channel. Set against the average contract value, it says whether a new client is profitable before they come back.

The number is fragile, and it should say so. Hires are few, so one contract more or less swings it. Hires lag the spend that earned them, so the current month always looks expensive. A contract that starts from a direct offer or a later post is not tied by Upwork to the proposal that opened the conversation, so a cost per hire from proposals can overstate the real price. An hourly contract has no value until hours are billed, so any revenue shown beside it is an estimate and should be marked as one.

Because of all this, cost per hire is a quarterly question and cost per reply is the weekly one. A falling cost per reply with a flat win rate usually means the replies are coming from the wrong clients.

How Upsight uses it. The P&L view shows acquisition cost beside the average deal and break-even, reads contracts from the agency's message threads with how each began, values Connects at the agency's measured top-up price before the list price, and marks hourly revenue as estimated. The consolidated funnel shows Connects per win under its stage cards, from the Connects ledger.