How do you prove a change in lead routing actually raised sales?

Why last year is the wrong comparison

Comparing this year with last year cannot separate a lead routing change from a new lead source, a different campaign mix, a price or offer change, a new team or a week of absences. Any of them can move sales more than routing does. Vendor results often compare different campaigns, different agents or different months, which has the same problem.

A control group fixes it by running the old way alongside the new one. Whatever else happens that month happens to both groups.

How to set up a fair test

1. Split at random, as leads arrive. Before anyone dials, so nobody can choose which leads go where.
2. Run both groups in the same weeks. Same agents, same dialer and CRM, same scripts and offers. Only the routing differs.
3. Score on sales per contact. The value of sales, by your floor’s own definition of a sale, divided by leads reached, each lead counted once however many calls it took. It counts both how often a lead reached buys and how much. Why conversion alone is not enough.
4. Keep every lead in its group. Whether or not the routing was followed on a given call, so the comparison cannot be flattered by dropping awkward leads.
5. Track adherence. How often the agent the routing chose was the agent who worked the lead. A low figure means the test measures less of the change than it could.

What to agree before the start

• The definition of a sale. Application taken, policy issued, or sale still active after cancellations. Write it down and use it for both groups.
• The decision rule. What result counts as a win, and how long the test runs before anyone decides.
• The rules on swaps and reassignment. How long an unworked lead stays with an agent, and whether agents may trade leads. These decide adherence.
• Which leads are reported separately. High-intent leads that buy from whoever answers dilute the gap. Keep them in their groups and report them on their own line.

What spoils a test

• Splitting by hand. If anyone chose which leads went where, the comparison is no longer fair.
• Sales recorded late, or cancelled weeks after the call, can make the latest weeks look worse than they are.
• Judging too early. The first weeks are noisy, and a single month, good or bad, says little about the next. Why a pilot can trail the control group at first.

This is the same method BestPair uses on collection floors, where the score is dollars collected per account contacted. On a sales floor, sales per contact takes its place.