Conversion

There is no universal conversion. A purchase is one. So is a form submission, a trial start, a phone call, a qualified lead, an app install, a subscription renewal. Each is a thing someone chose to count.

The problems all follow from that.

One team counts leads, another counts sales. Both call the number “conversions” and both are right inside their own report. Put the two figures side by side and the comparison is meaningless — this is easily the most common measurement argument in marketing, and it is definitional rather than technical.

The same event counted at different points. A trial start and a paid conversion are different events with wildly different values, and “conversions” in an ad platform is usually whichever one was configured as the optimisation target — often the earlier, more plentiful one, because it trains the platform faster.

A lead is not revenue. Counting leads as conversions makes any cost-per-conversion look good and says nothing about whether the money came back. If the two are not connected downstream, the number is a volume metric wearing an outcome’s name.

The practical discipline. Write down what counts, at what point in the journey, with what value attached, and keep that definition attached to the number wherever it travels. Most disputes about “our numbers don’t match” dissolve the moment both sides state their definition — and the ones that survive are usually about deduplication or attribution windows instead.