Position-based Attribution

Position-based attribution is a multi-touch model that assigns credit weight by each touch's position in the path; usually the first and last touches take the largest shares.

Position-based attribution answers the intuition that in most journeys two moments matter most: where the user is discovered and where they convert. Unlike linear, which weights all equally, and unlike time-decay, which weights only the end, this family highlights the first and last and gives the middle a smaller share. Its most common configuration is the U-shaped model with a 40-20-40 split (40% to the first touch, 40% to the last, 20% to the middle), but the ratios are not fixed, and that flexibility is both a strength and a hazard: each weighting is an assumption about the business, and chosen arbitrarily it merely moves the number. The model is honest when discovery and closing really are the two pillars of the journey; for journeys whose weight sits in the middle, it is a poor choice.