Self-reported attribution

Attribution is harder in leadgen than almost anywhere, because the journey from first touch to closed deal is long, multi-channel, and mostly invisible. A buyer hears about you on a podcast, sees a LinkedIn post, asks a peer, reads a few pages months apart, and finally types your name into Google and converts. The analytics stack records the last click - the branded search - and credits the channel that did the least. Everything that actually created the demand happened in the dark.

Self-reported attribution is the blunt fix: just ask. A single “how did you hear about us?” field on the form captures what the tracking can’t, because the buyer remembers the podcast even though no cookie does. It’s imprecise, it’s unfashionable next to a multi-touch model, and it’s frequently more accurate than the model, because it sees the channels that don’t leave a trackable click.

Why it works where tracking fails

The channels that drive demand generation - podcasts, communities, word of mouth, dark social - are exactly the ones analytics can’t see. They produce no click to attribute, so a tracking-based model systematically under-credits them and over-credits the last-click capture channels like branded search and retargeting. Self-reported data corrects the bias because it asks the one system that witnessed the whole journey: the buyer’s memory.

It pairs naturally with the technical side. Feeding closed-won outcomes back to the platforms fixes optimisation for the trackable channels; self-reported attribution catches the untrackable ones. Together they give a fuller picture than either the analytics stack or the ad platforms can alone.

Reading it honestly

• It’s directional, not precise. Memory is fuzzy, people misremember, and recency bias means they name the most recent touch. Use it to see which channels show up at all, not to allocate budget to two decimal places.
• Free-text versus options is a trade. A free-text box captures surprises but is a nightmare to analyse. A dropdown is clean but only ever shows you the channels you already thought to list, so it can’t reveal the one you’re missing. A hybrid - common options plus an “other” box - is the usual compromise.
• Aggregate it against the trackable data, don’t replace it. Self-reported attribution is a correction layer over the analytics, most useful where the two disagree. A channel that’s invisible to tracking but keeps showing up in self-reported answers is the one you’re under-investing in.