Reach and frequency in CTV are measured at the household level by default, not the person level. This is a fundamental difference from linear TV (which uses panel-based person-level GRP measurement) and from mobile digital (which tracks individual device IDs). A CTV impression delivered to a household could be seen by one person or four — the platform typically knows the device, not the individual viewer. This shapes how planners should interpret CTV reach numbers and why cross-platform reach deduplication remains an unsolved problem in India.
Household-level vs person-level reach
CTV platforms identify viewers through device IDs (the TV's advertising ID, assigned by the OS — Android TV, Tizen, webOS), household IPs, or registered account logins. JioHotstar, with 100M+ registered users, can link a CTV session to an account — getting closer to person-level measurement. Platforms without login data (some FAST channels, unregistered smart TV browsers) can only infer household identity from IP address.
The implication for reach planning: a CTV reach figure of 5 million households does not equal 5 million individuals. Average India household size is 4.4 people. Assuming each household contains at least one relevant viewer, the addressable person-reach is higher than the household figure suggests — but not all household members are co-viewing, and the platform cannot tell you which ones were present.
Frequency management in CTV
Frequency capping on CTV operates at the household device ID level. A cap of 3 impressions per household per day means the device (not a specific person) will not be served more than 3 ads from that campaign in 24 hours. The practical result:
- In single-person households, this works as intended
- In multi-person households, the cap may under-serve some viewers (if one person already hit the cap, another person using the same TV later in the day will not see the ad)
- Across multiple devices in the same household (TV + mobile), frequency is not deduplicated unless the DSP has household graph data linking the devices
DV360 supports household-level frequency management when publishers pass a consistent household ID in the bid request. For India CTV, major publishers (JioHotstar, SonyLIV) do pass household identifiers through Magnite and PubMatic deal flows. Open auction buys have less consistent household ID coverage.
Cross-platform reach deduplication in India
The most important unsolved measurement problem in India video planning: deduplicating reach across linear TV, CTV, and mobile video. A brand running a campaign across Star TV (linear), JioHotstar (CTV + mobile), and YouTube is reaching overlapping audiences that cannot currently be measured in a unified way.
The challenges are structural:
- BARC India measures linear TV via a panel of ~50,000 households. CTV is not in scope.
- JioHotstar reports digital impressions separately from linear Star TV ratings. The overlap between users who watch both is not published.
- There is no India equivalent of Nielsen ONE or VideoAmp that merges panel + digital data into deduplicated cross-platform reach curves.
For now, India planners should treat CTV reach as incremental to linear TV — assume minimal overlap with linear-TV-only viewers, and avoid adding CTV and linear GRPs as if they measure the same person-exposures. This is a conservative assumption that will become more precise as cross-platform measurement infrastructure develops.
Optimal frequency for CTV in India
Industry evidence from India CTV campaigns suggests:
- Brand awareness campaigns: 3–5 exposures per household per week is sufficient for recall lift. Above 7, recall plateaus and brand sentiment starts to decline.
- Performance campaigns (app install, purchase): Higher frequency (5–8/week) correlates with conversion lift, but the window is narrow — excessive frequency on CTV creates negative associations faster than on mobile.
- Live event campaigns (IPL, cricket): Frequency norms don't apply in the same way — viewers accept higher ad frequency during live events as part of the format. Publishers know this and price it accordingly.