attribution

Pick one source of truth for the conversion count, usually your CRM or backend where revenue is real, and de-duplicate platform claims against it. Never sum platform-reported conversions, since each platform claims credit for the same sales. Read directional trends rather than expecting absolute numbers to match.

Show first-touch and last-touch side by side rather than picking one, because the gap between them reveals how demand is created versus closed. For long B2B cycles, position-based attribution is a defensible primary model, supplemented by self-reported attribution at demo or signup.

Append the visitor's anonymous analytics ID to the outbound booking link via the tool's metadata passthrough, then merge it back in the webhook using an identify call with the booking email. Only ever pass the anonymous ID, never an identified email or UUID, to avoid corrupting user profiles.

Platform-reported ROAS is systematically inflated through self-crediting, view-through conversions, and generous attribution windows, so it is not incremental ROAS. Run an incrementality test such as a geo holdout or on-off test to measure the true lift, and use incremental CPA to drive budget decisions.

Direct traffic is mostly a measurement gap, not a real channel: it includes stripped referrers, dark social, app-to-web visits, and dropped tracking. A large direct share usually means your top-of-funnel is working but under-measured, so use self-reported surveys and segment branded versus non-branded search.

Use multi-touch attribution for day-to-day tactical optimization when you have clean user-level tracking. Use marketing mix modeling when spend is material across multiple channels including offline ones like podcasts or TV, since MMM works on aggregate data and sees channels tracking cannot.