Connected TV is not television, and buying it like television is expensive
It looks like TV, it is sold by people who used to sell TV, and it behaves like programmatic display with a very large screen attached.
iStudios, Media practice
Digital video, which includes connected TV, reached $78bn in the US in 2025 and grew 25.4%, one of the fastest of any major format. The budget arriving is largely coming from teams whose instincts were formed by linear television, and those instincts are a poor guide here.
Three things that are genuinely different
- Frequency is not managed for you. Linear had natural caps imposed by inventory. Connected TV has none, and across several apps and several supply paths the same household can be hit dozens of times a week without anyone intending it.
- The same impression is sold more than once. Supply path complexity means a single avail can reach you through multiple routes at different prices, and you can outbid yourself without noticing.
- Measurement is neither TV's nor digital's. Panel methods do not cover it well and digital attribution cannot see it at all, which leaves experiments and mix modelling doing the work.
The most common connected TV problem is not reach or price. It is that a small number of households absorbed most of the budget.
What to insist on
Household level frequency reporting across every supplier, not per platform. A deduplicated reach curve. Transparency on the supply path, and a willingness to consolidate it even when that means giving up some theoretical efficiency. Then a geo test, because none of the above tells you whether the channel produced anything.
Connected TV is a genuinely good channel. It is also the channel where the gap between the reported number and the real one is currently widest, and the advertisers doing well in it are the ones treating that as a media buying problem rather than a video production one.