Every CTV impression metric in India is a device-level count. Platforms report how many times an ad was served to a device — not how many people saw it. In a country where the average household has 4 members and the living room TV is shared, this creates a systematic undercount of actual reach that planners have not yet fully corrected for.
The measurement gap
A campaign that delivers 10 million device impressions in India's metro markets has almost certainly been seen by 20–30 million people. The gap between device impressions and person-level reach is the co-viewing gap — and it is currently invisible in standard campaign reporting.
This has two opposite effects depending on how you interpret the data:
- Reach is understated: If you compare CTV reach (device-level) to mobile OTT reach (individual-level), CTV looks like it reaches fewer people per rupee. Adjusting for co-viewing often reverses this — CTV may be reaching more people at equivalent or lower cost-per-person.
- Frequency is understated: If you cap frequency at 5 device impressions and each impression reaches 3 household members, those members have each seen the ad more than 5 times. The real person-level frequency is higher than your plan assumed.
Why co-viewing is hard to measure
The technical barriers to measuring co-viewing in India:
- No person-level detection on most devices: Smart TVs do not know how many people are in the room. The TV does not have a camera enabled for audience detection (even where cameras exist, privacy settings disable this). The device logs one session, regardless of audience size.
- No set-meter penetration at scale: BARC's traditional TV measurement uses physical meters on a panel of households that count viewers in the room. BARC Streaming is extending this methodology to OTT, but panel size and adoption are still growing — real-time co-viewing data is not available for most campaigns.
- Platform data is device-first: JioHotstar, SonyLIV, and other platforms build their reporting infrastructure around device/account sessions, not household members. Adding person-level tracking requires panel research, not just platform data.
- Household composition data is sparse: Even if co-viewing multipliers could be applied, applying the right multiplier (2x vs 4x) requires knowing the household size and composition for each impression — data that is not in most DSP profiles.
Tools and methods that help
Despite the gaps, practitioners use several approaches to estimate co-viewing:
- Panel-based co-viewing studies: Some platforms commission or share panel research showing average co-viewers per session by content genre and daypart. Use these to apply genre-specific multipliers (sports: 3–5x, drama: 2–3x, late night: 1–1.5x).
- Brand lift studies: Post-campaign brand lift research can compare exposed vs control household members. If the study recruits from the household rather than just the account holder, it captures co-viewing effects on brand metrics.
- BARC Streaming data: Where available, BARC Streaming provides panel-measured person-level data for participating platforms. This is the most reliable source of co-viewing-adjusted audience data in India — and coverage is expanding.
- Smart TV ACR data: Samsung Ads and LG Ad Solutions use Automatic Content Recognition to detect what is on screen and can correlate with viewing time. Some smart TV manufacturers are beginning to pilot ambient audience detection — but this is early-stage in India.
How to adjust your planning for co-viewing
Practical adjustments for campaigns running on India CTV:
- Apply a conservative 2x co-viewing multiplier when calculating cost-per-person reached. Use 2.5x for prime time or sports, 1.5x for late night or non-prime.
- When presenting CTV reach to clients, always footnote that reach figures are device-level and the actual person-level reach is estimated to be 2–3x higher.
- Set frequency caps at the household IP level, not device level, where your platform allows it. Target 3–4 household IP exposures per week rather than 5–7 device exposures.
- When running brand lift measurement, work with your research partner to recruit from the household rather than just the logged-in account holder.
- Use genre-based multipliers rather than a flat platform average — a sports buy during IPL has a very different co-viewing rate than a drama buy at the same time of day.
The honest limitation
Co-viewing multipliers in India are directional estimates, not measured data. The 2.5x figure commonly used in planning is an industry approximation — not a platform-verified, campaign-specific number. Apply it consistently and transparently, communicate its approximate nature to clients, and monitor for BARC Streaming data releases that will eventually provide panel-based grounding for these estimates.