Ignoring self-reported attribution because it isn't 'clean' data

Ignoring self-reported attribution because it isn't 'clean' data

Analytics-minded teams often discard the 'How did you hear about us?' field as messy and unscientific. In long-cycle B2B, that field frequently captures partner influence your tracking pixels structurally cannot.

Why pixels miss it:
— Multi-device, multi-stakeholder buying journeys break cookie chains. A champion researches on mobile, a procurement lead signs on a corporate VPN.
— Dark-social partner influence — a Slack community, a private webinar, a podcast mention — leaves no trackable click at all.
— Several attribution studies (e.g., work popularized by HockeyStack and Dreamdata) show self-reported sources surfacing partner channels that last-click ranks near zero.

The fix — triangulate, don't choose:
— Run self-reported attribution as a parallel directional layer, not a replacement for tracked data.
— Look for systematic divergence: channels that rank high in surveys but invisible in pixels are your dark-funnel partners.
— Reconcile quarterly; reallocate test budget toward the gaps.

Caveat: self-reported data carries recency and salience bias — buyers over-credit the last thing they remember. It is a compass, not a ruler.

Implications: a program that trusts only clickstream data will systematically defund its hardest-to-track, often highest-influence partners.