Streaming services defined themselves against advertising. No breaks, no interruptions, watch what you want. That was the proposition and for a decade it held.
Nearly all of them now sell an advertising-supported tier, and several push it as the default option. The reversal is complete and the reasoning behind it is not mysterious.
The arithmetic
The relevant number is revenue per subscriber, and the assumption everyone made was that a paying subscriber produces more than an advertising-supported one.
For a viewer who watches a lot, that is no longer reliably true. Advertising revenue scales with hours viewed. Subscription revenue does not — a heavy viewer and a light viewer pay the same.
Which means a heavy viewer on an advertising tier can generate more than they would on a subscription, particularly in markets with strong advertising rates.
The services have the data on how much each individual watches and can see this directly, which is presumably why the tiers appeared roughly simultaneously across the industry.
Price sensitivity was underestimated
The other half. Subscription prices rose steadily for years and the number of services people were expected to hold rose alongside.
At some point the total became more than households would pay, and cancellation rates rose.
A cheaper tier with advertising retains someone who would otherwise leave entirely. A subscriber generating modest advertising revenue is worth more than a former subscriber generating none.
Which reframes the tier as retention rather than as monetisation, and that is probably the more accurate way to read it.
What is different from broadcast advertising
Worth being precise, because the objection that this is a return to television is only partly right.
Targeting. Broadcast advertising is bought against a programme and an assumed audience. Streaming advertising is bought against an individual profile, which is worth substantially more per impression.
Measurement. The service knows whether the advert was played, whether the viewer was present, and in some cases what happened afterwards. Broadcast measurement is a sampled estimate.
Load. Advertising loads on streaming tiers are generally lighter than broadcast, at least currently, which is a deliberate positioning choice and has been creeping upward.
Frequency control. In principle, streaming can avoid showing the same advert repeatedly. In practice the inventory available to a given profile is often thin and people see the same thing many times, which is the most common complaint.
The consequences for what gets made
The part that interests me most and is least discussed.
Advertising revenue depends on hours viewed, which changes what a service wants from its catalogue.
Under a subscription model, a service wants things compelling enough to justify the subscription. Prestige, distinctiveness, something you cannot get elsewhere.
Under an advertising model, a service wants volume of viewing. Long-running series, large libraries of ordinary material, things people leave on.
These pull in different directions, and the shift toward volume programming and away from expensive limited series is visible across several services over the last few years.
Break structure also returns as a production consideration. Content made for uninterrupted viewing has no natural break points, and inserting them arbitrarily is noticeable. Content made with breaks in mind is structured differently, which is a real change to how episodes are written.
Where the tiers actually differ
Beyond the advertising, and this is worth checking before choosing.
Advertising tiers frequently have reduced resolution, no offline downloads, fewer simultaneous streams, and in some cases a smaller catalogue where certain titles are excluded for licensing reasons.
That last one is the least advertised and the most annoying. Paying less and discovering that specific things are unavailable is a different proposition from paying less and seeing adverts.
What I expect next
Advertising loads to increase gradually, because there is no competitive pressure keeping them low once everyone has done it.
The price gap between tiers to narrow, with advertising-free becoming a premium position at a higher price rather than the standard offering at the current one.
And more services making the advertising tier the default presented at signup, which several already do.
None of this is a conspiracy. It is what happens when an industry discovers that a business model it abandoned was better suited to its actual economics than the one it replaced it with, which has happened before in media and will happen again.
What it means for viewers, practically
A few conclusions I have drawn as somebody who pays for several of these.
Check what the cheaper tier actually removes before switching. Resolution and offline downloads are the two that catch people, and neither is prominently stated.
Assume the advertising load will rise. Choosing a tier on the basis of how light the breaks are today is choosing on a number that has no reason to stay where it is.
And audit what you hold, annually. The reason total household spending on this crept upward is that subscriptions renew silently and cancellation requires an action, which is the same asymmetry that autoplay exploits at a smaller scale.
The longer arc
What strikes me most is how closely the sector has converged on the structure it replaced. Bundles, tiers, advertising, and content commissioned for volume.
The technology changed completely and the economics arrived back at roughly the same place, which suggests the economics were never really about the technology.