Marketing Attribution Models Explained: First-Touch, Last-Click, Multi-Touch
Five models, five different answers to the same question. Here's what they actually measure and which to trust.
A visitor sees your tweet on Monday, reads a blog post linked from Hacker News on Wednesday, lands on your pricing page from a Google search on Friday, and signs up. The next morning the marketing team meets to ask: which of those channels gets the credit?
The answer depends entirely on which attribution model you use, and the same customer journey can produce wildly different conclusions — enough to change which channel you invest in next quarter. This article walks through the five models, what each one really measures, and which is sensible for a SaaS team.
What attribution actually is
Attribution is the rule for assigning credit when one conversion involves multiple touchpoints. It is not a measurement of reality — it is a chosen approximation. Every model is "wrong" in some way. The job is picking the one whose wrongness you can live with.
The five models, ranked by how often they show up
1. Last-click attribution
The default in most tools, including Google Analytics. The conversion is credited to whichever source brought the visitor on the session they actually converted in.
- What it captures well: Bottom-of-funnel sources. Direct, branded search, and the channels people use when they're already convinced.
- What it under-counts: Anything that introduced the customer to you in the first place. Twitter, podcasts, and content marketing get systematically ignored.
- When it lies most: When customers research for days and convert via direct/branded search. Last-click attributes everything to direct, when the actual win was the content piece they read three days earlier.
2. First-touch attribution
The opposite default — credit goes to whatever source first introduced the visitor to your site, regardless of what happened in the sessions afterward.
- What it captures well: Top-of-funnel sources. The channels that introduce people to your product get the credit.
- What it under-counts: Anything that closes the deal. Pricing pages, retargeting, demos, and any nurturing-stage content get nothing.
- When it lies most: When the first touch is a one-off blog post that didn't actually persuade anyone. A visitor who first saw you in a comparison article and then converted six weeks later via paid search makes the comparison article look magical.
3. Linear attribution
Equal credit split across every touchpoint in the journey. Four touches? Each gets 25%.
What it captures well: the breadth of the journey. No source is silently zeroed out.
What it gets wrong: a passing glance at a tweet gets the same weight as the demo that closed the deal. Reality isn't linear.
4. Time-decay attribution
Touchpoints closer to the conversion get more credit; earlier ones get less. Splits the difference between first-touch and last-click in a defensible way.
What it captures well: medium-funnel sources — comparison articles, pricing-page content, retargeting ads — that traditional models tend to under-credit.
What it gets wrong: long-cycle discoveries. The article that introduced the customer to your space three months ago gets a tiny sliver of credit even if it was the most important moment.
5. U-shaped (position-based) attribution
Splits credit between the first touch (40%), the last touch (40%), and everything in between (20%, divided equally). The model that says "introducing them and closing them are equally important; the middle is filler."
What it captures well: two-stage SaaS journeys. Discovery and conversion both matter, nurturing touches are real but secondary.
What it gets wrong: three-stage SaaS journeys. The genuinely important middle touch — say, a free trial — gets shortchanged.
Side-by-side: the same journey, five models
Imagine a customer journey: Twitter → Hacker News → Google Search → Direct → Convert. Here's how each model assigns credit to a $100 conversion.
| model | HN | Search | Direct | |
|---|---|---|---|---|
| Last-click | $0 | $0 | $0 | $100 |
| First-touch | $100 | $0 | $0 | $0 |
| Linear | $25 | $25 | $25 | $25 |
| Time-decay | $10 | $15 | $30 | $45 |
| U-shaped | $40 | $10 | $10 | $40 |
Same data, five different answers. If you spend $300/mo on Twitter content, last-click would tell you the channel is worthless; first-touch would tell you to double the budget. The truth is somewhere in between, and no model knows where.
What attribution can't do without cookies
Multi-touch attribution requires linking sessions across days and devices. That requires persistent identifiers, which is exactly what cookie-free analytics deliberately avoids — see Cookieless Tracking Explained.
The practical effect: in a cookie-free world, your attribution model usually collapses to session-level attribution — credit goes to whichever source brought the visitor on the session that converted. Cross-day journeys lose their early touches, which mostly become "direct."
Which model should you actually use?
For most indie SaaS, the answer is unsexier than the model comparison suggests:
// default
Session-level attribution + revenue per source
Look at which referrer source produced visits that converted in the same session. Don't fight the cookie-free reality — it forces you to focus on the touchpoint that actually closed the deal, which is usually what matters for budget allocation.
// when you scale
Add a self-reported attribution survey
Ask new customers "How did you hear about us?" with a free-text field. The data is messy but uniquely valuable — it captures cross-day, cross-device, cross-channel journeys that no automated model sees. Check it monthly.
// last resort
Build a multi-touch model only when channel decisions are six-figure decisions
If you're spending five figures a month per channel and need to defend allocation rigorously, time-decay or U-shaped multi-touch becomes worth the implementation cost. Below that, the model precision is noise.
The bottom line
Attribution models are decision tools, not truth tools. Pick the simplest one that gives you actionable answers, supplement it with a self-reported field for the journey gaps, and don't let model debates become a substitute for talking to the customers who actually paid you.
For most cookie-free SaaS analytics setups — including Datibase — the default is session-level attribution to the referrer source that closed the deal. Combined with disciplined UTM tagging, that single number answers most channel-budget questions you can act on.
See revenue by source, not just clicks
Datibase shows session-level attribution from referrer / UTM through to paid conversion — the simplest model that's actually actionable.
Try Datibase free