What counts as a sale when you match agents to leads?

Use the outcome the business is paid on

On an outbound sales floor, "a sale" can mean several different events, and they are not worth the same to the business. For matching agents to leads, use your floor’s own definition: the outcome the business is paid on. That is the number anyone will judge a routing change by, so it is the number the test should score.

The usual choices:

• A sale made on the call. The application taken, the order placed or the contract agreed while the agent is still on the line. It is known the same day.
• A sale that survives. Still on the books after cancellations, a cooling-off period or the first payment. Closer to revenue, and slower to count.
• A booked appointment. Where the floor sets meetings, tours or demos for someone else to close.
• An attended appointment. Where the business is paid only when the person turns up.

If the floor reports one of these and is paid on another, use the one it is paid on. A routing change that lifts sales made on the call, while the sales that survive stay flat, has not raised what the business earns.

Choose it before the test, and use it for both groups

Write the definition down before the first lead is split. Then apply it, unchanged, to the leads matched by BestPair and to the control group on your current routing, in the same weeks. The gap between the two groups only means something when both sides are counted by the same rule.

Settle these in writing:

1. Which outcome codes count as a sale, and which do not, such as a quote sent, a callback booked or an application started and left unfinished.
2. What happens to a cancelled sale. Whether it comes back out of the count, and up to when.
3. How a sale is valued. Revenue, premium, the fee the business earns per appointment, or a plain count where every sale is worth about the same.
4. Whether a sale to an existing customer counts the same as a sale to a new one.

Score it as sales per contact

With a sale defined, the result is sales per contact: the value of sales divided by leads reached, with each lead counted once however many calls it took. A lead reached means a real conversation with the person who can say yes, found from your own dialer or CRM outcome codes.

Sales per contact weighs how often a lead reached buys and how much it buys. A plain count of sales treats a small add-on and a large policy as one sale each, which hides a routing change that sends bigger sales to the agents who close them. What contact-to-sale conversion is, and why sales per contact is the better score.

Outcomes that settle later take longer to read

A sale made on the call is known when the call ends. A sale that has to survive a cooling-off period, or an appointment that has to be attended, is known only once that time has passed. The latest weeks of any test hold outcomes that have not settled yet, in both groups.

The stricter the definition, the longer you wait before a month can be read. You can watch the early outcome for movement and judge on the final one. The control group waits exactly as long as the matched group, so the comparison stays fair while you wait. We do not give a fixed window here, because it depends on how long your own outcomes take to settle. How to set up and read a fair test.