Connecting activity to money that arrived.

Leads is the wrong measure, and here is the proof

One campaign produces sixty leads. Another produces twenty, and more collected revenue. Judged on leads you cut the second one. Every business that measures marketing on lead volume is making some version of this mistake, and the only fix is to carry the chain further.

Carry it all the way to collected

Won is not revenue. Invoiced is not revenue. Collected is revenue. The difference between them is usually small and occasionally catastrophic, and the channels that produce customers who do not pay are exactly the channels that look best on a won-deals report.

First and last touch are both wrong

First touch flatters advertising; last touch flatters whoever sent the final email. Keep the whole chain and look at it more than one way — the conclusions that survive both readings are the ones worth acting on.

Say what you cannot trace

Somebody who saw an advert, thought about it for three weeks and then typed your name into a search engine is not reliably traceable. Showing that share as untraceable is more useful than distributing it across channels to make the totals balance — and a model that always reconciles perfectly is a model doing exactly that.

Work backwards from a payment

Start at an invoice that was paid and walk back: which conversation, which channel, which campaign. This direction is far more useful than working forwards from spend, and almost no tool supports it.

Per customer, never per send

Four hundred calls made is a cost. Money collected divided by customers won, per channel, is the number that changes a decision. Include real costs where you know them — telephony and model usage are knowable; guessing at salary is not.

• Meetings held, per channel
• Customers won, per channel
• Money collected, per channel
• Cost, where it is reliably known

Expect it to tell you something you dislike

Usually that human outbound produces the fewest meetings and the most revenue per customer, which is an argument for putting people on the largest opportunities and agents on everything else. An attribution model that only confirms what you already believed is not measuring.