Attribution honesty

Two numbers show up on the front of this site. One says the demand generation function I built at Forbes went from $0 to eight figures in marketing-influenced revenue. The other says $72K of paid budget turned into $275K of direct-booked revenue in 2026. They sit in different cells of the proof grid, and when I talk about them they sit in different sentences. That is deliberate, and it took a few years of quarterly business reviews to learn why it matters.

Influenced revenue is a wide net. It is counted through Salesforce contact-role attribution: if a person we brought in through a form, an event, or a nurture is attached to an opportunity that later closes, the revenue counts as influenced. A seller worked the deal. Finance forecasted it. Marketing touched it somewhere along the way. The number is real, it is tracked conservatively, and it is the right way to describe what lifecycle programs do across 2019 to 2026, because lifecycle work is mostly about being present at the right moments of a long sales cycle.

Direct-booked revenue is a narrow net. It counts revenue where the paid campaign was the source of the booking and there is no other credible path: the account came in through the campaign, and the deal closed on that thread. It is smaller by design. In 2026 it was $275K on $72K of spend across LinkedIn, Google, Meta, and programmatic, with sourced pipeline sitting on top that has not closed yet.

The trouble starts when the two get blended. "Marketing drove $5.25M this year" reads well in a deck until someone in the room asks how, and the honest answer is that most of it was influence on deals sales was already working. Then the big number looks inflated, and the $275K, which was always clean, gets doubted along with it. One vague number poisons the specific number next to it. I have watched this happen to other teams and I have watched it nearly happen to mine, which is why every number on this site carries the method it was measured with, in small type, under the figure.

Keeping the sentences apart also keeps me honest about what each number can predict. The direct number is, roughly, a function of paid spend. Put more budget in and more bookings come out, until the audience saturates. The small model on the home page does exactly that arithmetic, and it says on its face that it is linear on 2026's ratio with no diminishing returns modeled, because that is all it is. The influenced number does not work that way. It moves with how many buying moments the lifecycle programs are present for, which depends on the sales calendar, the content, and the size of the opportunity pipeline far more than on any budget line. Modeling it off spend would be pretending.

There is a version of this that sounds like modesty. It is closer to the opposite. When the two numbers are kept apart, the direct one becomes something I can defend line by line in front of a CFO, and the influenced one becomes a claim about the machine rather than a claim about me. Both come out stronger.

The working rules, for anyone building the same kind of reporting: report influenced and direct on separate lines with separate methods. Never put "over" in front of a number the source does not carry. When the attribution model changes, rebase the history and say the numbers are post-rebase (the 2026 paid figures above are). And when a number cannot be traced to a source, cut it, even when it was a good number.