Most ad platforms default to last-click attribution: whoever touched the conversion last gets 100% credit. That flatters brand search and starves prospecting. Multi-touch attribution (MTA) distributes credit across the path — but only if you can see the path.
A realistic B2B / FinServ journey
−12d organic/search /blog/ecsp-guide −9d paid/meta prospecting reel −3d paid/google brand search −1d direct /signup now conversion lead logged
Last-click says “direct.” Multi-touch says Meta and organic assisted; Google captured intent; direct closed. Budget decisions change when you see the chain.
Common MTA models (and trade-offs)
- Linear — equal credit to each touch. Simple, often too flat.
- Time decay — more credit to recent touches. Good for short cycles.
- Position-based (U-shaped) — 40% first, 40% last, 20% middle. Popular for considered purchases.
- Data-driven — platform ML (Google DDA, Meta). Better than last-click, still platform-biased.
Operator teams pick a model that matches sales cycle length, then reconcile with first-party logs — not the other way around.
Assisted conversions you should track
Report assisted rate by channel: “% of conversions where this channel appeared earlier in the path.” High assist + low last-click on Meta prospecting often means do not cut prospecting when brand search ROAS looks perfect.
Identity is the hard part
Cross-device and cross-domain identity requires a stable first-party ID, click ID capture on landing, and consent where GDPR applies. Without that, MTA collapses to guesswork.
We log the nexus on our stack — every touchpoint, every click ID — then decide budget moves. Platform dashboards alone are not enough.