Self-Reported Attribution: The Most Honest Signal You Have

Self-reported attribution is the practice of asking buyers directly how they first heard about you, usually with a 'how did you hear about us' field on a high-intent form, and using that answer as a primary attribution signal. It sounds primitive next to a multi-touch model, and that is exactly why it works: it captures the podcast, the Slack community, the colleague's recommendation, and the conference hallway conversation that no pixel will ever see.

Pixels only record touches that happen on properties you control or can tag. The majority of B2B demand is created in places you cannot instrument, so pixel-based attribution is precise about the small slice it sees and blind to the large slice that actually drove the deal. A human typing I heard you on the SaaS podcast is messy data, but it is messy data about the truth, which beats clean data about a fraction of it.

The standard objection is that buyers misremember or give vague answers, and they do; you will get Google from someone who Googled your name after a referral. But you are not relying on any single response. Across hundreds of answers, patterns emerge that no individual recall error can hide: if podcast keeps appearing, the podcast is working, even if the exact episode is fuzzy. Aggregate self-report is far more reliable than individual self-report.

The second objection is that free text is hard to analyze. That used to be true; now you pipe the answers into your warehouse and use an LLM to normalize 'saw your founder on LI,' 'LinkedIn post,' and 'someone shared your link' into one bucket. The grind of categorizing thousands of open-ended answers, which once killed self-report programs, is now a cheap automated step. The only real requirement is making the field mandatory on the forms that matter.

Self-report is not a replacement for tracking; it is the second leg of a triangle. Put the self-reported channel next to the tracked first-touch and the multi-touch model for the same deals and read the disagreements. When self-report says podcast but tracking shows branded search, you have just proven the podcast created the demand that search merely harvested. That single insight reallocates budget more honestly than any model alone.

Stored on the contact record in HubSpot or Salesforce and joined on your identity graph, self-report becomes a queryable signal, not a quarterly survey gathering dust. You can segment closed-won revenue by self-reported source and compare it to spend per channel, exposing the gap between what you can track and what actually works. The channels that win on self-report but vanish from pixels are precisely the ones agencies tell you to cut.

The practical fear is that an extra field hurts form conversion. Mitigate it: put the question only on high-intent forms like demo requests where intent is already high, use a short open-text box rather than a long dropdown, and never gate the rest of the form on a perfect answer. A modest dip on a demo form is a trade worth making for honest attribution on your most valuable conversions.

Then close the loop. Route the answers into the warehouse, normalize them with an automated job, and build one report: self-reported source by closed-won revenue, side by side with multi-touch and spend. Review it monthly alongside your other lenses. The point is not to crown self-report as the only truth, but to give the untrackable majority of your demand a seat at the table it was previously denied.