Attribution models, explained without the vendor jargon

Last-click attribution gives 100% of the credit to the final touchpoint before conversion. It's simple and easy to explain, but it systematically undercounts channels that open a journey rather than close it — brand search, early content, and awareness advertising all look weaker than they are.

Linear attribution splits credit evenly across every touchpoint. It's fairer to top-of-funnel channels but treats a brand search click the same as a passing display impression, which usually overcorrects in the other direction.

Time-decay attribution weights credit toward touchpoints closer to conversion, which suits longer sales cycles reasonably well, but still relies on someone deciding the decay rate rather than deriving it from actual behaviour.

Data-driven attribution, where available, uses your own conversion data to algorithmically weight touchpoints based on their actual observed contribution. It's the most defensible model where you have enough conversion volume to make the algorithm reliable — typically not a fit for low-volume accounts, where a simpler model is both more transparent and more stable.