How do you actually measure lead quality by source instead of arguing about it?

Quality is not an opinion once you pick the measures

Arguments about lead quality persist because each side cites a different metric. Marketing points at volume and cost. Operations points at the calls that went nowhere. Neither is lying; they are measuring different stages.

Fix it by agreeing on a fixed set of outcome measures per source, computed the same way every period. The argument then becomes a conversation about what to do, which is the correct argument to be having.

The five measures worth tracking

• Booking rate on real opportunities. Not on all contacts — a source that sends spam should be penalized in the junk rate, not in the booking rate.
• Average and median ticket. Median matters because one large job can hide a source full of small ones.
• Cancellation and no-show rate. Some sources book easily and hold poorly.
• Callback and warranty rate. Sources that skew toward price-driven customers often skew toward rework.
• Junk share. Wrong numbers, robocalls, out of area, services you do not offer, and existing customers miscounted as new.

The junk bucket is where most of the disagreement lives

Until the non-opportunities are separated out, every downstream rate is polluted, and a source with heavy junk will look like a source with a handling problem. Separating them by hand does not scale past a small sample.

This is a natural application for call analysis: classify every inbound call by whether it was a real service opportunity, then compute the rates on that clean denominator. The classification is auditable — each call keeps the evidence for why it was categorized the way it was.

Give the table time before you act on it

Booking rates on small samples swing hard. A source with a couple of dozen contacts in a month can look excellent or terrible for reasons that are pure chance. Set a minimum volume before a source's rates appear in the decision table, and let low-volume sources accumulate across a longer window.

Then act on the composite, not on any single measure. A source with a slightly worse booking rate and a much better average ticket is usually the better source, and that judgment only appears when the measures sit side by side on the same report.

Refresh the table on a fixed schedule and keep the history. Quality by source drifts as competitors enter, platforms change matching, and your own service area shifts, so last year's ranking is not evidence about this quarter.