Measurement ยท Attribution Models

Multi-touch attribution for CTV in India: how it works and where it breaks

Multi-touch attribution (MTA) attempts to assign conversion credit across multiple touchpoints in a customer's journey — rather than crediting only the last click or first impression. In web advertising, MTA models distribute credit across display, search, social, and email touchpoints. For CTV, MTA integration is possible in theory but faces structural constraints in India that make it harder than for web channels. Understanding what MTA can and cannot do for India CTV campaigns prevents misallocated credit and bad budget decisions.

Why CTV is hard to integrate into MTA

Standard MTA models in web advertising work because every channel touches a common identity layer — browser cookies or mobile app IDs — allowing a single user journey to be reconstructed across channels. CTV breaks this in two ways:

Device ID gap: CTV device IDs (GAID on Android TV, TIFA on Samsung) are different identity namespaces from mobile cookies and browser cookies. Stitching a CTV impression into a web or mobile conversion journey requires a cross-device identity bridge — a device graph that links the CTV device ID to the web or mobile identity that eventually converted. In India, this linkage succeeds for 20–40% of CTV impressions, leaving the majority unmapped.

Non-clickable inventory: CTV ads do not generate clicks. MTA models built primarily on click data (which most India performance measurement systems still are) will systematically ignore CTV's contribution entirely, or assign it near-zero weight in the model.

MTA model types and CTV compatibility

MTA modelCTV compatibilityIndia limitation
Last-clickNone — CTV cannot be the last clickSystematically credits performance channels; ignores CTV
First-touchPartial — if CTV is the first exposure with linked identityRequires identity match; 20–40% coverage
Linear (equal credit)Partial — credit distributed to all touchpoints including CTV where matchedOnly matched CTV impressions receive credit
Data-driven (algorithmic)Partial — model learns CTV's contribution from matched pathsBiased toward channels with full identity coverage
Geo holdout / incrementalityFull — does not require identity matchingRequires campaign scale; city-level conversion data

Practical MTA approach for India CTV

For advertisers running integrated digital and CTV campaigns in India, the recommended measurement approach is a layered one:

MTA for the connected portion: Run standard MTA across all digital channels (social, search, display, mobile video). For the CTV impressions that successfully resolve to device graph links, include them in the MTA model. This captures credit for the 20–40% of CTV impressions with resolvable identity.

Holdout test for the full CTV effect: Run a geo holdout test in parallel during major campaign flights to measure CTV's total incremental contribution — including the 60–80% of impressions that MTA cannot see. Use holdout results to calibrate the "true" CTV contribution that MTA underestimates.

Media mix modelling (MMM) for budget allocation: MMM uses aggregate sales and spend data rather than individual impression paths, avoiding the identity matching problem entirely. For brands running large enough CTV budgets to generate a detectable sales signal, MMM is a more reliable CTV attribution tool than path-level MTA in India.

The India MTA vendor landscape for CTV

Few India-specific MTA vendors have fully integrated CTV. AppsFlyer and Adjust offer cross-device attribution that includes some CTV-to-mobile path coverage. Global MTA vendors (Neustar, Nielsen, TransUnion) have limited India CTV identity graph depth. Advertisers building MTA models in India should verify which CTV publishers and identity namespaces each vendor covers before assuming full CTV integration.