Last-click attribution credits the final touchpoint before conversion, multi-touch attribution splits credit across every touchpoint in the journey, and closed-loop attribution ties credit to actual closed CRM revenue rather than a conversion event — agencies default to last-click because it’s simplest to set up, but it systematically misattributes upper-funnel channels and overstates the value of bottom-funnel and branded touchpoints.
Three attribution models, side by side
Every attribution model is answering the same underlying question — which marketing touchpoint deserves credit for a conversion — but each one draws the line for “credit” and “conversion” in a different place.
| Last-click | Multi-touch | Closed-loop | |
|---|---|---|---|
| What it measures | Credit to the final click/touch before conversion | Credit split across all touchpoints in the path | Actual closed revenue traced to source campaign |
| Pros | Simple, built into every ad platform, zero setup | Reflects full customer journey, fairer to upper-funnel | Ties spend to real revenue, not assumed lead value |
| Cons | Ignores upper-funnel touches, inflates branded/retargeting | Complex to build correctly, models can disagree | Requires CRM integration and longer feedback loop |
| Data requirements | Pixel or click ID, platform-native | Cross-channel tracking, ID resolution, a modeling layer | CRM with deal-to-campaign linking, sales-stage tracking |
| Best for | Fast-moving, single-channel, low-consideration purchases | Multi-channel campaigns with a documented longer journey | Lead-gen and B2B with a real sales process and CRM |
Last-click attribution
Last-click attribution gives 100% of the credit for a conversion to the final touchpoint a person interacted with before converting — the last ad they clicked, or in some setups, the last ad they clicked within that platform’s own attribution window. It is the default reporting model inside Meta Ads Manager, Google Ads, and most other platforms, which is exactly why it’s the model most agencies end up reporting on, even when they know it’s imperfect.
Why agencies default to it
- Zero setup — every ad platform reports it natively.
- Easy for clients to understand: one click, one conversion.
- Works reasonably well for single-channel, low-consideration purchases where there’s rarely more than one touchpoint anyway.
Where it misattributes
Last-click systematically overweights the touchpoint closest to conversion — branded search, retargeting, email — and underweights the touchpoints that actually created demand, like prospecting campaigns, top-of-funnel video, or organic social. A prospect who discovers a brand through a cold Meta ad, researches for two weeks, then converts through a branded Google search gives 100% of the attribution credit to the search click that intercepted demand the Meta ad originally created.
Multi-touch attribution
Multi-touch attribution distributes conversion credit across multiple touchpoints in a customer’s path, using a weighting rule — even distribution across all touches (linear), heavier weight on the first and last touch (U-shaped), increasing weight toward the touch closest to conversion (time-decay), or a statistically modeled weighting (algorithmic/data-driven).
Multi-touch is a real improvement over last-click for multi-channel campaigns because it stops treating every conversion as if it happened in a single click. The tradeoff is implementation complexity: it requires reliable cross-channel identity resolution (the same person needs to be recognized across a Meta ad, a Google ad, and a direct site visit), and different multi-touch models can produce meaningfully different credit splits for the same underlying data, which makes the output harder to explain to a client than a single last-click number.
Closed-loop attribution
Closed-loop attribution answers a different question than the other two. Instead of asking “which touchpoint gets credit for this conversion event,” it asks “what actually happened to this lead, and how much is it really worth.” It connects a lead’s source campaign to its final CRM outcome — closed-won, closed-lost, or still open — and reports ROAS based on real revenue rather than an assumed conversion value.
This makes closed-loop attribution the most accurate model for businesses with a real sales process, because it’s the only one of the three that accounts for lead quality. A last-click or multi-touch model will happily assign full credit to a campaign that generates 100 cheap, unqualified leads — closed-loop attribution will show that campaign’s true ROAS is near zero once none of those leads close.
When the CRM-integration effort of closed-loop is worth it
Closed-loop attribution is the most accurate model but also the most expensive to set up — it requires CRM integration, consistent UTM discipline, and deal-to-campaign linking maintained over time. That effort isn’t justified for every account. A practical framework:
Closed-loop is worth it when:
- The client has a real sales process — leads go through qualification, not an instant checkout.
- Deal values vary significantly between leads (a $500 job and a $15,000 job both start as “a lead”).
- Sales cycles run long enough that platform-reported conversion value is clearly a guess, not a fact.
- The client already has, or is willing to adopt, a CRM the agency can integrate with.
- Ad spend is high enough that a ROAS miscalculation has real budget consequences.
Last-click (or simple multi-touch) is enough when:
- The purchase is a single-step, low-consideration ecommerce transaction with consistent order values.
- There’s no CRM, and building one solely for attribution isn’t worth the client’s investment.
- Spend is small enough that the cost of building closed-loop tracking would exceed the value of the extra accuracy.
In practice, most agencies running lead-generation accounts for home services, B2B, healthcare, legal, financial services, or any vertical with a real sales conversation land firmly in the “closed-loop is worth it” category — the gap between platform-reported and true ROAS in those verticals is usually too large to report around. Platforms like Loomstrat exist specifically to make that integration lift smaller: leads sync from ad platforms into a built-in CRM, deals track through to closed-won automatically, and true ROAS shows up next to platform ROAS in the same dashboard, without a custom data-engineering project behind it.