The number. The report tabulates whom connected-television viewers in urban India watch with; its own survey covered 3,000 of them. On the smartphone, 72 percent watch alone. On the connected television, 19 percent. The rest of the television figure is the household, in the report’s five categories: 25 percent with a spouse or partner, 21 percent with the whole family, 17 percent with a sibling, 17 percent with parents or elders. On the phone, those four together come to 26 percent.

The source. The Connected TV Era: India’s New Prime Time, published by WPP Media and The Trade Desk and executed by Ormax Media, 19 August 2026. A dedicated survey of 3,000 connected-television viewers in urban India, May to June 2026, combined with the Ormax OTT Audience Report 2026 for the size of the audience. The report is gated; I read all 29 pages, and the table is on page 13. It is survey response, urban only, and the report says so itself: it “does not represent a census of the entire population”. Its own reading of the table is that one impression on the big screen reaches about 2.5 people, in a home of 3.3. The report’s count of connected-television households is modelled by WPP Media rather than counted. Nothing in the 29 pages measures a title: no completion, no drop-off, and no record of who in the room chose what played.

What it says. [inference] The two screens may be measuring two different things. A phone is usually one person, so what it records is likely to be what that person chose and how far they went. A connected television is usually a room, so what it records may be what the room sat through, chosen by somebody in it. If that is right, a completion number read off a television account could be the room’s tolerance rather than one viewer’s choice, and the person who picked the title need not be the person who stayed for the finale. On the report’s own numbers, the 19 percent would be the share of television viewing where the chooser and the viewer are the same person; on the phone that share would be 72 percent. The report does not ask who chose, so this is a reading of the table, not a result in it.

Where the data ends. Four things this number cannot do. It is urban India only, and it is what people say about whom they watch with, not a measurement of sessions. It records presence, never choice: the report never asks who held the remote. It says nothing about completion on either screen, because the report has no title-level data at all. And it cannot tell you which screen your own show is being watched on; [inference] that sits in your platform’s data, and I know of no public source for it. What the report does show is where the growth is. Ormax’s own annual series, which the report draws on for its audience size, runs 69.7 million in 2024, 129.2 million in 2025 and 207 million in 2026, and the report has rural connected-television audiences up 110 percent in a year. [inference] The share of Indian streaming that is watched by a room rather than a person is rising, and the measurement problem is rising with it.

What I would decide anyway. The measurement is completion split by device: for your last ten titles, the share of starters who reached the finale on phone accounts and on connected-television accounts, side by side. Every platform can produce it. I know of none that publishes it. The decision it changes is what a completion number is allowed to mean. If the reading above holds, a title that finishes strongly on phones may have been chosen and finished by the same person, which would make it a verdict on the story. A title that finishes strongly on televisions may have held a room, which would be a verdict on how it plays in company, a different thing and a rarer one. The phone curve could then be the one to read when the question is whether to buy the next season from this writer, and the television curve the one to read when the question is what goes into the 8 to 10 in the evening slot, where the report has connected television 78 percent active on a weekday, against 63 percent for linear. None of this can be checked against the report, because the report has no title-level data. It can be checked against your own. And if your platform cannot split completion by device today, that is the first thing to fix, because until it can, every completion number you sign against may be an average of a person and a room.