Attribution models
First-touch, last-touch, and linear — the three v1.0 models, in depth, with worked examples.
This page is the reference. For a more decision-oriented read, see the guide Picking the right attribution model.
The three v1.0 models
Follow The Spend ships with three attribution models. All three are live simultaneously — switch with one click in Pulse, no recalculation lag.
First-touch
Definition: The visitor's first session in their journey gets 100% of the credit for any conversion that follows.
Mental model: First-touch rewards the channel that introduced this person to your brand. Useful for evaluating top-of-funnel acquisition.
Last-touch
Definition: The visitor's most recent session before the conversion gets 100% of the credit.
Mental model: Last-touch rewards the channel that closed the sale. Useful for evaluating bottom-of-funnel performance — retargeting, abandoned-cart emails, late-stage paid ads.
Linear
Definition: Every session in the journey gets an equal share of the credit. A 4-session journey splits credit 25/25/25/25.
Mental model: Linear avoids over-rewarding any single touch. Useful for understanding your full marketing mix when journeys are genuinely multi-touch.
A worked example
Same visitor, same journey, three different attributions:
Day 1, 14:32 — Meta ad → /products/red-shoes (3 pages, 4 min)
Day 3, 09:15 — Klaviyo email → /collections/sale (2 pages, 1 min)
Day 5, 21:08 — Direct (typed) → /checkout → ORDER ($120)| Model | Meta ad credit | Klaviyo email credit | Direct credit |
|---|---|---|---|
| First-touch | $120 | $0 | $0 |
| Last-touch | $0 | $0 | $120 |
| Linear | $40 | $40 | $40 |
The journey is identical. Three models, three different stories about who deserves credit. None of them are "wrong" — they answer different questions.
How conversions are tied to journeys
For a session to count in attribution, two things must be true:
- The session must belong to a visitor we can link to the order. This is the visitor cookie ↔ Shopify customer ID link, set at checkout or login.
- The session must fall within the lookback window. Sessions older than the window don't count, even if they're in the same visitor's history.
The lookback window in v1.0 is 30 days, applied uniformly across all three models. Sessions before the 30-day boundary are not credited. (Configurable lookbacks are post-v1.)
What you cannot configure in v1.0
To set expectations clearly:
- No custom models. Position-based (40/20/40), time-decay, data-driven — all post-v1.
- No model weights or splits. Linear is fixed at equal weights.
- No per-channel overrides. You can't tell FTS "always credit Meta over email" or similar.
- No lookback window adjustment. 30 days, fixed.
Each of these is a real product decision and we'll take them up once v1.0 customer signal tells us which model nuance matters most.
Switching models in Pulse
The model toggle lives in the top-right of the channel performance table. Click First / Last / Linear to swap. Every panel in Pulse re-attributes instantly using the same underlying journey data.
Visitor Journeys also show the per-model credit on each session — you can see "this session contributed X under first-touch, Y under last-touch" without leaving the page.
How to think about which model to use
Match the model to your current question, not your strategic preference:
| Today's question | Use this model |
|---|---|
| "Should I increase spend on this awareness campaign?" | First-touch |
| "Is my retargeting actually closing sales?" | Last-touch |
| "Which channels in my mix are pulling weight?" | Linear |
| "Is this content piece bringing in new customers?" | First-touch |
| "Is my abandoned-cart email driving revenue?" | Last-touch |
| "How does my full marketing mix compare?" | Linear |
The single most valuable habit: switch models when the question changes. A weekly channel review for budget allocation might use first-touch + last-touch comparisons. A campaign post-mortem might use linear. Don't pick once and forget.
What each model under-credits
Every model has a known blind spot:
- First-touch under-credits closers. Channels that finish the sale (retargeting, email) look weaker than they really are.
- Last-touch under-credits acquirers. Top-of-funnel channels (awareness ads, content, organic search) look weaker than they really are. This is the model most ad platforms use internally, which is why your Meta dashboard often shows higher numbers than reality.
- Linear under-credits any genuinely-decisive touch. If a visitor truly was about to bounce until that one Klaviyo email pulled them back, linear gives that email the same 25% credit as a low-impact Direct session.
Knowing the blind spot is half the battle. The right move is usually to read the same data through two models and take the picture from both.
Where to go next
- Picking the right attribution model — decision-oriented walkthrough
- How attribution is computed — the technical detail
- Channels and sources — what each session's channel actually means