Attribution

Definition

Attribution is the model used to assign credit for a conversion across the multiple touchpoints (ads, emails, organic visits, referrals) a buyer interacted with before converting.

Almost no real conversions come from a single touchpoint. A B2B buyer might see your LinkedIn ad, return via Google search a week later, read three articles, then convert from a retargeting email. Attribution is how you decide which of those touchpoints get credit — and therefore where you allocate next quarter's spend.

There is no 'correct' attribution model. Each model encodes a different assumption about how influence works. Last-touch over-credits the bottom of the funnel; first-touch over-credits awareness; data-driven uses ML but is opaque. Most teams should view 2–3 models side by side rather than pick one.

Origin

Direct-response marketers in the 1980s used per-channel coupon codes as crude attribution. Digital attribution formalised in the early 2010s as multi-touch journeys became the norm and Google Analytics added attribution reports.

How it works

Define what counts as a conversion (the event being attributed).
Pick an attribution window (e.g. 30 days for B2B, 7 days for ecommerce).
Pick a model: last-touch, first-touch, linear, time-decay, position-based, or data-driven.
Compare two models side-by-side. The gap tells you what each model is over- or under-crediting.
Tie attribution to spend reallocation — otherwise it's a vanity exercise.

When to use it

Use when

For monthly or quarterly channel-mix decisions.
When stakeholders disagree about which channel is 'really' driving growth.
After a big spend change, to see how attribution shifts.

Skip when

Daily — attribution is too noisy at short horizons.
As the only input to spend decisions. Pair with incrementality testing.

Key metrics

Attributed revenue per channel.
Attributed cost per acquisition (CAC) by channel.
Path length (touchpoints to convert).
Time to convert (days from first touch to conversion).

Examples

Last-touch credited paid search with 70% of revenue; data-driven dropped that to 41%.
Without attribution, every channel claims to be the one that drove the deal.
Attribution showed our LinkedIn ads were a top-of-funnel assist, not a closer.

In practice at Makreate

Makreate marketing engagements report on multiple attribution models in parallel — last-touch, first-touch, and data-driven — so spend decisions are grounded in pattern, not one model's bias. A recent fintech client was about to cut LinkedIn ads because last-touch credited only 8% of conversions. Multi-touch analysis showed LinkedIn was the first touch on 31% of conversions and an assist on another 22%. Cutting it would have cost more than it saved.

Advertising →

Common mistakes

Picking a single attribution model and treating its output as truth.
Using a window that's too short for your sales cycle. B2B with 30-day windows misses real influence.
Not running incrementality tests to validate attribution claims.
Letting each ad platform self-attribute without cross-platform deduplication.

Frequently asked

What's the best attribution model?

There isn't one. Last-touch is simplest, data-driven is most rigorous, and pattern-spotting across 2–3 models beats any single model.

How do iOS privacy changes affect attribution?

Significantly. Apple's ATT and SKAdNetwork limit cross-app tracking. Server-side tracking, first-party data and incrementality testing have become more important.

How does attribution differ from incrementality?

Attribution divides credit across observed touchpoints. Incrementality measures whether spending on a channel actually causes more conversions — usually via a holdout test.