Measuring Marketing When Half the Leads Call You

Every business wants to know which marketing works. Most measurement advice assumes a clean digital funnel that ends in an online purchase. For the large number of businesses where the enquiry is a phone call and the sale happens weeks later in a meeting, that advice does not transfer.

Accept that attribution is approximate

Someone sees a Meta advertisement, searches your name two weeks later, reads three pages, asks a colleague, then calls. No attribution model represents that accurately, and the ones claiming to are making choices they do not show you.

The useful goal is not precision. It is a consistent measure, applied the same way every month, that is good enough to decide where the next increment of budget goes. If two channels differ by a factor of two, your measurement is good enough to act on.

Measure what you can measure properly

Form submissions, call clicks, WhatsApp clicks, email clicks — these are all trackable with modest effort and cover most enquiry routes. Add call tracking where call volume justifies the cost.

Do not let the existence of untrackable paths become a reason to track nothing. Partial measurement applied consistently is far more useful than none, and the gap narrows as you improve it.

Ask, because people will tell you

'How did you find us?' asked by the person taking the call, recorded consistently, produces data no analytics tool can. It is imprecise — people misremember and give the last touchpoint — but it captures word of mouth, offline exposure and referrals that are otherwise invisible.

The key is consistency. Asked every time and recorded in the same place, it becomes a usable series. Asked occasionally, it is anecdote.

Close the loop on outcomes

Record which enquiries became customers, and what they were worth. Even a spreadsheet with source, date, outcome and value transforms what you can conclude.

This is where cost per enquiry becomes cost per customer, and the two rankings are frequently different. A channel producing expensive enquiries that close at three times the rate is the cheaper channel, and only outcome data reveals that.

Use holdouts when attribution fails entirely

For channels that resist attribution — brand advertising, offline, awareness campaigns — the practical test is turning them off in one region or for one period and observing what happens to total enquiries.

It is blunt and it works, because it measures the thing you actually care about: whether total business changes. Many businesses are reluctant to try it, and it is often the only honest answer available.

Report the same way every month

Consistency of method matters more than sophistication. A simple report — enquiries by source, cost per enquiry, cost per customer where known, produced identically each month — supports better decisions than an elaborate model that changes each time someone improves it.

And include what did not work. A report showing only improvement is not a report of reality, and everyone reading it knows that.