Attribution in advertising is the process of assigning credit for a conversion or outcome to one or more ad exposures that preceded it. In digital advertising, this is usually done with cookies — a user clicks a display ad, a cookie is set, and when they convert, the cookie fires and credits the ad. CTV has no clicks and no cookies. Attribution for connected TV therefore uses different mechanisms: view-through windows, IP matching, household graphs, and statistical modelling. Each method answers a slightly different question about campaign impact and comes with different accuracy trade-offs.
What CTV attribution measures
CTV attribution measures the relationship between ad exposure on a TV screen and a downstream action — typically one of:
- App install or open: Did households that saw the CTV ad install or reopen the advertiser's app at higher rates than unexposed households?
- Website visit: Did CTV-exposed households visit the advertiser's website more than a control group?
- Brand metric lift: Did awareness, recall, or purchase intent improve among CTV-exposed survey respondents vs a control?
- Sales or revenue lift: Did regions, households, or cohorts exposed to CTV ads show measurable incremental sales vs matched unexposed groups?
None of these are direct click-to-conversion measurements. CTV attribution is always inferential — it estimates impact rather than measuring a direct causal chain from impression to conversion.
View-through attribution: the CTV default
View-through attribution (VTA) credits a conversion to an ad impression if the conversion occurs within a specified window after the impression, even without a click. For CTV, a common window is 24 hours for app installs and 7 days for website visits. Any conversion within that window from a device connected to the same household IP as the CTV impression is attributed to the campaign.
VTA is the default in DSP reporting (DV360, The Trade Desk) because it requires no additional setup. Its limitation: it can over-attribute. A viewer who would have installed the app anyway — because they saw a search ad or heard about the brand on social media — gets counted as a CTV-driven install if they happen to be in the same household as a CTV impression. The inflate-and-celebrate problem is real in CTV VTA reporting, particularly in India where household sizes are large (4+ people) and IP-based matching produces household-level rather than person-level attribution.
The four CTV attribution approaches
- IP/household matching: Match CTV impression IPs to downstream conversion events. The most common, least accurate approach. Works directionally for comparing relative campaign performance. Accuracy degrades with dynamic IPs and CGNAT. See CTV attribution in India for India-specific limitations.
- Brand lift studies: Survey-based measurement of brand metric changes (awareness, recall, purchase intent) between CTV-exposed and unexposed groups. Measures brand impact accurately but not conversion. Requires a measurement vendor (Kantar, Nielsen) and minimum spend.
- Incrementality testing: Controlled experiment holding out a matched audience segment from CTV ads and measuring the difference in outcomes vs the exposed group. The gold standard for measuring true causal lift. See incrementality testing in CTV.
- Media mix modelling: Statistical model estimating CTV's contribution to business outcomes from aggregate spend and sales data. Requires 12+ months of consistent data. See media mix modelling and CTV.
Choosing the right method for India
The method choice depends on campaign objective and budget:
- Brand campaign, Rs 25L–1Cr budget: Brand lift study via Kantar or Nielsen. VTA from DSP as secondary directional signal. Do not use VTA as the primary metric for brand campaign evaluation.
- Performance campaign (app install, purchase), Rs 25L–1Cr: IP-based VTA via DSP + MMP integration (AppsFlyer, Adjust). Treat numbers as directional. Run an incrementality test on 20% of budget if campaign is recurring.
- Large ongoing spend (Rs 1Cr+ annualised): MMM as the long-term effectiveness framework. Quarterly incrementality tests. Brand lift on major campaign flights. VTA as operational reporting only, not strategic evaluation.