How do you measure reach in India CTV campaigns?

In a standard programmatic CTV campaign report from DV360 or The Trade Desk, "reach" means the number of unique device IDs that received at least one impression. This is device-level reach, not viewer-level reach. A CTV device (smart TV, streaming stick) is a shared household device — the number of actual people who saw the ad is likely 2–2.5 times higher than the device reach count, based on India's average co-viewing multiplier. When a publisher reports "15 million viewers reached," they may be applying a co-viewing multiplier to a smaller device ID count — these two numbers are not directly comparable.

Cross-publisher deduplication in India CTV is modelled, not measured, because there is no common identity layer connecting device IDs across JioHotstar, Zee5, SonyLIV, and other publishers. When running a multi-platform CTV campaign, each publisher or DSP reports its own reach number. Adding these together double-counts households reached on multiple platforms. Industry rule of thumb: assume 25–40% audience overlap when running on 2–3 major India AVOD publishers simultaneously. A combined reach of 30 million claimed by three publishers likely deduplicates to 18–22 million unique households. Third-party measurement vendors like DoubleVerify can provide independent reach verification, but cross-publisher deduplication requires integration with all publishers in the campaign — which is not always possible, particularly for JioHotstar direct inventory.

Use both, reported separately with their basis clearly stated. For media planning and campaign goal-setting, use device reach as the base metric (the number you can actually control and measure) and then apply a co-viewing multiplier appropriate to the content environment — 2.0–2.5x for general entertainment, 3.0–4.5x for live cricket. Report device reach as the verified number and viewer reach as the estimated number. Do not add co-viewing-adjusted publisher reach numbers to device-level DSP reach numbers — they measure different things and cannot be combined without normalisation.