Multi-touch attribution distributes credit for an iGaming conversion across every marketing touchpoint in the player's journey using a defined weighting model.
Multi-Touch Attribution (MTA)
**TL;DR:** Multi-touch attribution distributes credit for an iGaming conversion across every marketing touchpoint in the player's journey using a defined weighting model.
What it means
Where last-click awards 100% to the final touch, MTA splits credit across multiple steps — first touch, mid funnel, last touch — using models like linear, time-decay, position-based, or data-driven (Markov chains, Shapley value). For iGaming, MTA tries to answer: "How much of this FTD do I owe the TV spot vs the affiliate review vs the Google brand ad?"
MTA usually requires deterministic identity (logged-in user, device graph) plus media data piped into a CDP or attribution tool (e.g. Adjust, AppsFlyer, Northbeam, Rockerbox).
Formula / How it's measured
Not a single formula — depends on the model. Example using linear MTA across 4 touchpoints (TV → Meta → Affiliate review → Google brand) on a $300 FTD value: each touch gets 25% credit, so each is credited $75 of revenue and a 0.25 conversion.
A time-decay model would weight more heavily towards Google brand and affiliate; a U-shaped model heavily weights first and last touch.
Why it matters for operators
MTA helps operators avoid over-funding brand search and retargeting at the expense of demand-creating channels. Combined with incrementality testing, it produces a more honest channel ROI picture and informs media-mix decisions ahead of state launches or sport seasons.
Common benchmarks (2026)
- Maturity: ~30% of mid-large iGaming brands run any MTA; <10% trust it as the primary model
- Tools used: Adjust, AppsFlyer, Northbeam, Rockerbox, in-house BigQuery models
- Typical re-weighting impact: brand search credit drops 20–40%, paid social rises 15–25%
- US sportsbook operators investing most heavily due to TV spend
- Data freshness: 24h–72h vs real-time last-click
Common mistakes
- Trusting deterministic-only MTA in iOS/cookieless environments (huge data gaps)
- Comparing MTA outputs across vendors without normalising models
- Using MTA without incrementality validation — it shows correlation, not causation
See also