First touch vs last touch vs multi touch attribution: how each model works, and which to use
First-touch attribution gives all the credit to the customer's first tracked interaction, last-touch to the final one, and multi-touch splits it across several. First touch suits awareness analysis, last touch conversion analysis, and multi touch longer paths.
Each model moves credit toward a predictable set of channels: the choice changes how channels compare, not how many conversions there were.
First-touch attribution
- How it works: the whole conversion goes to the first tracked interaction, such as an ad click.
- Best for: seeing which channels introduce new customers and generate demand.
- Drawback: emails, retargeting and the channels that close the sale get no credit.
- Tracking risk: the first interaction is less likely to be tracked on long or cross-device paths.
Last-touch attribution
- How it works: the whole conversion goes to the final tracked interaction before purchase.
- Default: long the default in many tools; Google Analytics 4 and Google Ads now default to data-driven attribution.
- Best for: seeing which channels close sales, and optimising the final funnel steps.
- Drawback: awareness gets no credit; brand search and retargeting get credit for customers already decided.
Multi-touch attribution and its variants
- Linear splits credit evenly across every interaction.
- Time decay gives more to interactions closer to the purchase.
- Position-based (U-shaped) gives a large share to the first and last, and splits the rest.
- Data-driven learns weights from converting and non-converting paths; it needs a large volume of conversions.
- Best for longer paths; drawback: needs reliable tracking across the whole path.
The three compared
Example path: a social ad click, a blog visit from search, an email, then a purchase after a brand search click.
- First touch: all to the social ad. Last touch: all to brand search.
- Linear: a quarter each. Position-based: typically 40% each to social and brand search, 10% each to the rest.
- Stability: single-touch models are simple and stable; multi-touch weights can shift as paths change.
What all three have in common
- Same total: all three distribute the same tracked conversions.
- Untracked conversions (in-store, connected TV or radio, untracked devices) are credited to nothing.
- Sales that would have happened anyway are still credited to an interaction on the path.
- So no attribution model can report that a channel added nothing.
How to see what a switch would change
- Export the same period under both models; the difference by channel is the size of the decision.
- Usual pattern: brand search and retargeting lose share to earlier interactions; broad-reach social and display gain.
- If not, paths are shorter than expected, or early interactions go untracked.
When each model is the better choice
- First touch: acquisition questions.
- Last touch: conversion questions; also the easiest to keep consistent.
- Position-based or data-driven: short, well-tracked digital paths with several steps, typical of ecommerce.
- B2B and long purchases: much of the path goes untracked, so single-touch models are easier to keep honest.
- Budget split across channels: contribution measurement, usually a marketing mix model.
What attribution models cannot tell you
- No attribution model measures what a channel caused: each distributes credit among interactions on converting paths.
- Offline channels are not covered comparably to digital ones.
- For those questions: a marketing mix model for contribution, experiments for one channel's direct effect.
How to choose
- Short, fully tracked digital path: data-driven if your platform supports it, position-based if not.
- Long or partly offline path: the simplest model you can explain, usually last touch.
- Budget across channels: contribution measurement calibrated with experiments.
Cassandra is an AI marketing brain for growth teams. It measures each channel with a Bayesian marketing mix model checked by live geo experiments, and does the work within the team's rules. Start for free with your website and data.
Changing attribution model
- The cost is comparability: year-on-year figures, benchmarks and targets do not carry over.
- Run both models in parallel for a full reporting cycle before switching.
- Show readers the side-by-side period, so a falling share is not read as a performance problem.
Questions
What are the key differences between first click and last click attribution?
First click gives all the credit to the first tracked interaction, so it favours channels that introduce customers. Last click gives it all to the final interaction, so it favours channels that close sales.
What are the four types of attribution?
The most common are first touch, last touch, linear and time decay. Position-based and data-driven are also widely used. All but the two single-touch models are forms of multi-touch attribution.
What is the drawback to using the last touch attribution model?
It gives no credit to anything before the final interaction, so awareness activity looks unproductive, and channels close to the purchase get credit for customers who had already decided to buy.
Which attribution model is most accurate?
All of them divide the same tracked conversions, so none measures what a channel caused. Data-driven attribution divides the credit more defensibly than a fixed formula on well-tracked paths.
How often should we change attribution model?
As rarely as possible. Every historical comparison breaks on the day the model changes, and consistency over time is usually worth more than a marginally better model.