Lead attribution explained: first-touch, last-touch and multi-touch

A buyer in 2026 rarely meets you once. They might find you through a Google search, read a blog post a week later, click a retargeting ad the next day, and finally fill in a funnel after a recommendation. Which of those four touches "caused" the lead? Attribution is the set of rules you use to answer that, and the answer changes which campaigns you fund.

The three models, in plain terms

First-touch attribution

All the credit goes to the first interaction. If the customer's first touch was the organic search, the search gets 100% of the credit regardless of what happened after.

Credit to: first channel the lead ever touched
Example path: Search → Blog → Retarget ad → Funnel
Credit:       100% Search

When it fits: when you are measuring top-of-funnel demand generation — brand, SEO, PR. First-touch tells you which channels originate relationships, which matters if your problem is "nobody knows we exist".

Last-touch attribution

All the credit goes to the final interaction before the conversion. In the example above, the funnel (or the ad that delivered them to it) gets 100%.

Credit to: last channel before conversion
Example path: Search → Blog → Retarget ad → Funnel
Credit:       100% Funnel (or its source ad)

When it fits: when you are measuring bottom-of-funnel efficiency — which closing channel actually produces the enquiry. Most default platform reports (Google Ads "conversions", Meta "results") are last-touch, which is why they all claim to be your best performer.

Multi-touch attribution

Credit is spread across several or all touches using a rule — linear (equal split), time-decay (more to recent touches), position-based (40% first, 40% last, 20% middle), or data-driven (modelled on actual conversion patterns).

Linear example (4 touches): 25% each
Time-decay example:          more weight to Retarget ad & Funnel
Position-based example:      40% Search, 40% Funnel, 20% split of the rest

When it fits: when you have a real sequence of touches and enough volume to make the split meaningful. Multi-touch is the most honest model, and the most expensive to get right.

A worked example

Suppose a lead arrives via this path and the business spends £300 on the search campaign, £150 on the blog (content), and £200 on retargeting, with the funnel itself being free:

Note the totals always sum to the same £650 of attributed spend — attribution only moves credit around, it does not create or destroy it. The decision changes completely: first-touch says "fund search", last-touch says "fund retargeting", and position-based says "search and retargeting both matter, content a little".

Which model should a small business use?

For most small businesses, the honest answer is: start with last-touch for closing efficiency and first-touch for demand, and only graduate to multi-touch once you have the volume and the discipline to use it.

• Under ~200 conversions a month: multi-touch data-driven models are statistically thin. You will be fitting noise. Use last-touch to manage your closing channels and first-touch to judge your awareness spend, and keep the two separate.
• 200–1,000 conversions a month: linear or position-based multi-touch becomes defensible. The patterns are stable enough to act on.
• Over 1,000 a month: consider data-driven attribution, but only if your tracking is clean (see limits below).

The trap is using one model for everything. A business that manages its SEO on last-touch data will conclude SEO does nothing, because SEO is almost always a first or middle touch. It gets no credit and gets defunded — then the top of the funnel dries up and last-touch channels have nothing to close.

The honest limits of attribution

Attribution is a model of reality, not reality. Plan around these limits or you will over-trust the numbers.

• The missing touches. A recommendation over the phone, a trade-show conversation, or a word-of-mouth chat leaves no click. Any model that only sees digital touches systematically under-credits offline and over-credits the last ad.
• Cross-device and cross-browser gaps. A user researches on a phone, converts on a laptop, in private mode. The stitching breaks and the lead looks "direct" or unattributed.
• The equal-credit illusion. Linear multi-touch implies every touch contributed equally, which is rarely true. A blog post read once six weeks ago is not worth the same as the retargeting ad that caught them the day they converted.
• Attribution does not equal causation. Just because a channel was last does not mean it caused the conversion. The buyer may have already decided and merely clicked the final link.
• Platform bias. Each ad platform's native attribution is built to make that platform look good. Comparing Meta's last-touch claim against Google's last-touch claim is comparing two self-interested stories.

The practical mitigation is to anchor on a number attribution cannot distort: cost per qualified lead, by source, measured against actual revenue. See our ROI and attribution overview for how that is wired up. Because our tracking pixel ships underneath the funnel builder, the source is captured at the point of capture rather than reconstructed after the fact — which removes a whole class of stitching errors.

Common mistakes

• Trusting one platform's dashboard as truth. It is optimised to retain your spend. Reconcile against your own records.
• Switching models without annotation. If you move from last-touch to linear, your numbers shift even if nothing in the business changed. Label the change or you will chase ghosts.
• Attributing to the funnel page, not the source. The funnel is the conversion point, not the credit-worthy channel. Credit the ad or search that delivered the visitor.
• Ignoring view-through. Impressions with no click still influence behaviour. A model with zero view-through credit will under-value brand and display.

Putting attribution to work

For a small business, the goal is not a perfect model — it is a defensible one you actually use. Pick last-touch for closing channels, first-touch for demand, and report cost per qualified lead by source as the tie-breaker. When you are ready for multi-touch, start linear and only move to position-based or data-driven once the volume supports it.

If you want the infrastructure without a tagging project, the funnel builder captures source automatically, and pricing is flat per lead with unlimited users so your whole team can see the same numbers. And for a quick sanity check on the economics, our cost-per-lead calculator turns spend, qualification and close rates into CPQL and ROAS instantly.

All four of the models above — first-touch, last-touch, linear and position-based — are worked out for every lead as it converts, so you can switch between them on the lead page and see the same journey argued four ways rather than committing to one and rebuilding your reports later. Time-decay and data-driven models are not built. How attribution works in Leadtrak.

Attribution will never be exact. The win is being roughly right and consistent, rather than precisely wrong in a way that funds the wrong channel.