Self-Reported Attribution
- Self-reported attribution asks the buyer directly — the complement to tracked attribution.
- It sees what analytics cannot: AI answers, word of mouth, and other untracked influence.
- In the AI era it is the direct read on the invisible shortlist — necessary, not optional.
- Treat it as triangulation: divergence from tracked source and referral traffic is the signal.
Consensus definition
Tracked attribution follows clicks through cookies, parameters, and referrers; self-reported attribution instead asks the buyer to name the channel that brought them. The two answer different questions and are known to disagree, because tracked models systematically miss touchpoints they cannot instrument — dark social, word of mouth, offline, and now AI answers consumed without a click (zero-click).1 Self-reported data is noisy and biased — recency effects, recall error, last-thing-remembered — but it captures something tracking cannot: the buyer's conscious account of what influenced them, including the contacts that left no digital trace.
rhinegold operator refinement
Rhinegold's position: in an AI-era funnel, self-reported attribution moves from nice-to-have to necessary, because the most important new touchpoint — being named or recommended in an AI answer — is precisely the one tracked attribution is worst at seeing. It is the direct counterpart to the invisible shortlist: when a buyer arrives already knowing the brand and analytics records a brand-search or direct visit, only the buyer can tell you an AI put you on the list. The discipline is to treat self-reported data as triangulation, not truth — read it alongside referral traffic and tracked attribution, and use divergence between the three as the signal, since each is blind where the others see.
Operational use
Add a single "How did you hear about us?" question to high-intent forms (demo, contact, quote), with an open field plus a short option list that explicitly includes AI assistants. Read it in aggregate as a trend, cross-tabulated against tracked source, and watch the share that credits AI or "a recommendation" against the share analytics labels direct/brand — the gap is the invisible influence.
Measurement boundary
Self-reported data is biased and imprecise: buyers over-credit the last or most memorable touch, under-report channels they barely registered, and answer inconsistently. It cannot allocate fractional credit across a path the way modeled attribution attempts to, and small-sample categories are volatile. It is a directional cross-check, not a system of record — its value is precisely where tracking is blind, not as a replacement for it.
Distinct from
Against tracked/modeled Attribution, which reconstructs the path from instrumented signals — self-reported asks the human instead, and sees the untracked. Against LLM Referral Traffic, which counts the clicks AI actually sent; self-reported also captures the influence that produced no click at all. Against a Counterfactual, which estimates incrementality — self-reported describes perceived influence, which is not the same as causal lift.
Obstacles & resolutions
Operational note
Common mistakes
- Treating self-reported answers as precise channel allocation — they are a biased, directional account, not a ledger.
- Offering only fixed options with no AI/recommendation choice and no open field — the new touchpoint then cannot be reported at all.
- Reading single responses instead of aggregate trends and cross-tabs against tracked source, where the real signal lives.
Where consensus is missing
There is no standard question wording, option set, or method for reconciling self-reported data with tracked attribution — practitioners blend them ad hoc. How much to trust self-reported AI-influence figures, given recall bias, is genuinely unsettled, and no benchmark exists for the expected tracked-vs-reported gap.
Sources & deeper reading
- rhinegold Insights, Episode 04 — How to Measure What You Can't See
- rhinegold self-reported-vs-tracked triangulation methodology — survey design and gap analysis for AI-era funnels, shared with engaged clients and partners.