Solution Finder
What is your customer trying to achieve? Pick the problem closest to what they actually said — this translates it into the right methodology, data requirements, and Lab.
You said: “Understand which digital touchpoints contributed to conversion”
Multi-Touch Attribution
MeasurementWhy this fits
Multi-Touch Attribution allocates conversion credit across the individual touchpoints in a customer's journey.
What it answers
Across the multiple ads and channels a customer interacts with before converting, which touchpoints deserve credit, and how should that credit be split?
How it works
Individual customer journeys (sequences of impressions and clicks across channels) are reconstructed, then a credit-allocation rule — last click, first touch, linear, or position-based — assigns a share of the conversion to each touchpoint in the journey.
Data required
- Individual-level or device-level touchpoint data across channels
- Timestamped sequence of impressions/clicks per user journey
- Conversion event data linked back to the journey
- Consistent identity resolution across touchpoints (increasingly difficult with privacy changes)
Discovery questions
Can touchpoints actually be stitched into a single customer journey?
MTA depends on identity resolution across channels, which has become significantly harder as third-party cookies and device identifiers are restricted.
Which attribution rule does the business currently use, and why?
Different rules produce very different channel-level credit, and switching rules can dramatically change which channels look 'effective.'
Is this meant to replace or complement an MMM view?
MTA and MMM answer different questions and are often most useful together, not as substitutes.
Implementation approaches
Risks & limitations
Risks
- Heavy reliance on individual-level identity data that is increasingly restricted by privacy changes
- Choice of attribution rule can be gamed or misused to make a preferred channel look better
Limitations
- Cannot capture the effect of channels with weak digital identity signals (e.g. TV, OOH)
- Rule-based models are heuristics, not causal estimates — they describe correlation in the journey, not what caused the conversion