CTV attribution in India is a genuinely difficult problem. CTV is a lean-back, non-click medium — there is no click-through to track, no cookie to follow across sessions, and no standard integration between a viewer's TV and their purchase behaviour. The attribution gap is real: a brand runs a CTV campaign, sees strong VCR and reach numbers, but cannot directly tie those impressions to app installs, website visits, or store purchases. Most India CTV campaigns solve this imperfectly, with a patchwork of IP matching, brand lift studies, and media mix modelling. Understanding what each method measures — and what it cannot — is the foundation of any honest CTV measurement framework.
Why CTV attribution is harder than digital
Three structural features of CTV make attribution harder than display or mobile:
- No click: CTV ads are non-interactive in the standard sense. There is no URL to click, no pixel to fire on engagement, and no direct session handoff between the TV and a downstream conversion action. The viewer watches the ad and then — if it worked — picks up their phone or opens a laptop to act.
- Device discontinuity: The TV where the ad was seen and the device where the conversion happens are almost always different. Connecting a CTV impression to a mobile app install or web purchase requires cross-device identity matching — which in India relies primarily on IP address, since cookie-based matching and device graphs are less developed than in the US or UK.
- No cookie: CTV devices don't support cookies. The device advertising ID (IFA) is available for frequency capping within a campaign but is not usable for attribution to downstream web/app conversions without a DSP or MMP integration that can match the IFA to a downstream device ID.
Attribution methods available in India
- Brand lift studies (Kantar, Nielsen): Survey-based. Expose a test group to CTV ads and measure brand awareness, ad recall, purchase intent lift vs a matched control group. Measures brand impact, not conversion. Appropriate for brand campaigns. Requires minimum spend (typically Rs 50L+) and 2–4 week campaign duration. Standard for large FMCG and auto brands.
- IP-based attribution: The most common performance attribution method in India CTV. Match the IP address of households that saw CTV ads to the IP addresses of subsequent website visitors or app sessions. Accuracy degrades when households share IPs across many devices or use mobile data rather than home WiFi for downstream actions. Offered by DSPs (DV360, TTD) and MMPs (AppsFlyer, Adjust).
- Media mix modelling (MMM): Statistical regression modelling that estimates the contribution of CTV spend to business outcomes alongside other media. Does not require impression-level matching — it works from aggregate spend and outcome data. Requires 12+ months of data to be reliable. Used by large advertisers (Hindustan Unilever, Maruti) who run consistent CTV spend.
- Incrementality testing: Holdout test. Run CTV ads in some regions or audience segments and withhold from matched controls. Measure lift in outcomes (app installs, sales, website visits) between test and control. The most rigorous method but requires campaign budget large enough to support meaningful holdout sizes.
- ACR-based attribution: Automatic Content Recognition data from smart TVs can identify which households saw a specific ad, then match to purchase data panels. Limited availability in India — Samsung and LG have ACR capabilities but ACR-to-purchase matching for India is nascent.
IP-based attribution: the India default
IP-based attribution is what most India CTV campaigns use in practice, because it requires no special measurement vendor integration and is reported natively by DSPs. The process: the DSP records the household IP of each served CTV impression. When a web visit or app open is tracked by the advertiser's analytics or MMP, the IP is matched to the CTV impression log. Matches within a defined post-view window (typically 24 hours for app installs, 7–14 days for website visits) are credited as CTV-influenced conversions.
Limitations specific to India:
- Many India households use Jio or BSNL residential broadband with dynamic IPs that change daily. A household that saw a CTV ad on Monday may have a different IP on Tuesday when they visit the website — the match fails.
- Jio and Airtel mobile broadband uses CGNAT (carrier-grade NAT) which shares a single public IP across thousands of mobile subscribers. A CTV household IP that appears in the attribution match may map to thousands of Jio mobile users, producing false positives.
- The conversion action often happens on mobile data, not the same home WiFi network the TV is on — so the IPs don't match even if the person is the same.
For India CTV, IP-based attribution should be treated as directional, not precise. It underestimates true CTV influence (due to IP mismatch) and occasionally over-credits it (due to CGNAT false matches). Use it to compare relative CTV performance across campaigns and publishers, not as an absolute conversion count.
What India brands actually use
In practice, India CTV attribution breaks down by advertiser type:
- Large FMCG/auto brands (HUL, Maruti, Asian Paints): Brand lift studies as primary measurement. MMM for long-term effectiveness tracking. CTV treated as a brand medium — outcomes measured in brand metrics, not direct response.
- E-commerce and fintech (Flipkart, Zomato, PhonePe): IP-based attribution via AppsFlyer or Adjust MMP integration with DSP. Incrementality tests run quarterly. CTV measured on influenced install and reactivation, not last-touch.
- D2C brands (mid-tier spend): Platform-reported data + rough IP attribution. Most lack the budget for brand lift studies or proper MMM. Google's Reach Planner and DV360's built-in attribution is the practical ceiling for this segment.