Attribution Basics: Agreeing What Counts as a Conversion Before You Spend a Dollar
The single most common cause of a marketing report nobody trusts is not the channel mix. It is that "conversion" was never actually defined.
Almost every marketing report that gets quietly distrusted has the same root cause, and it is rarely the one people argue about in the room. It is not the channel mix, the creative, or even the spend level. It is that nobody wrote down, in advance, exactly what counts as a conversion — so every stakeholder is silently using a slightly different definition, and the report cannot satisfy all of them at once.
Four questions to settle before the first dollar is spent
- What is the conversion event, precisely? A form submit, a booked call, or a closed deal are three different funnels with three different timelines.
- What is the attribution window? A seven-day click window and a ninety-day view window will show the same campaign performing wildly differently.
- What happens to a conversion touched by three channels? Last-click, first-click and linear models will each hand credit to a different one.
- What is the reporting lag? A B2B sales cycle that takes six weeks means this month’s spend cannot honestly be judged against this month’s closed revenue.
None of these questions has a universally correct answer. A DTC brand with a same-day purchase and a B2B firm with a six-week sales cycle should not use the same model. The failure is not picking the "wrong" model — it is not picking one out loud, so the marketing team quietly uses one and the finance team quietly uses another, and they meet each month to argue about numbers that were never going to agree.
Write it down where everyone can see it
A single paragraph, agreed before spend starts: "A conversion is a booked scoping call. We attribute on a thirty-day click window, last non-direct click. We report on a two-week lag to account for the sales cycle." Four sentences, and most of the monthly argument disappears — not because the numbers changed, but because everyone is now arguing about the same number.
What a trustworthy report says about the bad months
A report that is green every month is not a good report — it is a report nobody is checking against reality. The agreement above only earns trust if the reporting team is willing to say, in a month a channel underperforms, that it underperformed, against the definition everyone signed off on, and say what changes next month as a result.
The argument about whether marketing "worked" is almost always an argument about a definition nobody agreed on. Settle the definition first and the argument mostly evaporates.