1. First-Touch
How it works: 100% of the credit goes to the first interaction.
When it wins: You are trying to measure demand-creation channels like cold LinkedIn Ads, content SEO for top-of-funnel queries, or podcast appearances. First-touch rewards the channel that created the user.
When it lies: You think first-touch reflects all the work it took to close the deal. It does not. It ignores everything after the first click — including sales, nurture, case studies, pricing page revisits. Use first-touch for demand-creation analysis, not for credit assignment of closed revenue.
2. Last-Touch
How it works: 100% of the credit goes to the final interaction before conversion.
When it wins: Short, simple funnels with one dominant conversion event. DR-heavy e-commerce below $500k revenue. Transactional info products with a single ad-to-purchase flow.
When it lies: Almost everywhere else. For B2B SaaS, last-touch over-credits branded search, retargeting, and "direct" traffic — systematically hiding the channels that originated demand. This is the single most damaging default in marketing analytics.
3. Linear
How it works: Equal credit to every touchpoint.
When it wins: Short funnels (2–5 touches) where touchpoints have comparable cost and no single one is dominant. Small info product funnels, early-stage SaaS.
When it lies: Long B2B cycles with dominant touchpoints. Retargeting-heavy funnels (cheap touches dilute expensive ones). Branded search after paid demand gen. See my deep dive on linear attribution.
4. Time Decay
How it works: More credit to touches closer to conversion. Typical half-life: 7 days.
When it wins: Sales-led motions where later touches (SDR calls, demo, pricing pages, case studies) genuinely do more work than earlier ones. Works for mid-market B2B SaaS with 30–90 day cycles.
When it lies: Top-of-funnel matters. Time decay under-credits demand creation by design — if the LinkedIn Ad that created the user was 90 days before conversion, it gets almost nothing. For brands that depend on cold demand, this is a slow bleed.
5. Position-Based (U-Shaped)
How it works: 40% to first touch, 40% to last touch, 20% split across middle touches.
When it wins: B2B SaaS with a cold-to-close flow where both demand creation AND closing touches matter. My default recommendation for most B2B SaaS operators under $10M ARR.
When it lies: Your funnel has more than 2 inflection points. For example, if there is a critical "SAL" milestone in the middle that represents sales qualification, U-shaped under-credits it.
6. W-Shaped
How it works: 30% to first touch, 30% to lead capture (form fill or MQL), 30% to SAL (sales qualified), 10% split across remaining touches.
When it wins: B2B SaaS with a clear funnel: inbound → MQL → SAL → closed. Matches the reality of a sales-led motion better than any simpler model. Best for $1M–$10M ARR teams with defined pipeline stages.
When it lies: PLG or self-serve motions where lead capture and SAL are the same event (a signup). W-shaped collapses into something simpler and loses its advantage.
7. Data-Driven
How it works: An algorithm (Markov chain, Shapley value, or proprietary ML) determines credit weights based on your conversion data. GA4 defaults to this. HockeyStack and Dreamdata offer versions.
When it wins: High conversion volume (500+ events per month per channel). Enterprise B2B or high-velocity DTC. The algorithm needs signal; at volume, it beats any rule-based model.
When it lies: Low-volume funnels. Under 100 conversions per month, the algorithm over-fits to noise. Numbers will shift dramatically between reporting periods without any real change in the business. If you are under $2M ARR, data-driven is a distraction.
Decision Matrix
| Your Situation | Use | Skip |
|---|---|---|
| B2B SaaS, $1M–$10M ARR, mixed inbound | U-shaped (default) or W-shaped | Last-touch, data-driven (too little data) |
| B2B SaaS, $10M+ ARR | Data-driven + W-shaped comparison | First-touch alone |
| DTC, $500k–$5M | U-shaped or time decay | Linear (biased by retargeting) |
| DTC, $10M+ | Data-driven + MMM | Any single rule-based |
| Info product, 7–14 day funnel | Linear or time decay | First-touch |
| Heavy demand-creation measurement | First-touch (as secondary model) | Last-touch, time decay |
| Heavy DR / conversion measurement | Last-touch (as secondary model) | First-touch |
The best operators look at 2–3 models simultaneously. If U-shaped and last-touch agree on channel rankings, you can trust the answer. When they disagree dramatically, that is the interesting case — investigate what is actually happening, then pick the model that matches your funnel shape.