Two series can be finished on the same day and released on entirely different schedules. The choice is made by the service, and it is about the shape of a subscription month.
A subscription is billed monthly
Streaming revenue arrives in monthly increments, so the operative question is not whether a viewer watches but whether they are still subscribed when the next charge lands.
A series delivered in one drop can be consumed inside a weekend. The remaining weeks of that billing cycle then have to be filled by something else or the subscription looks expendable.
A weekly release stretches a single title across a whole cycle and often into the next one. The same amount of content is doing considerably more retention work.
Conversation runs on a different clock
Weekly releases create a shared position in a story. Everyone watching is at the same point at the same time, which sustains discussion, speculation and clips between episodes.
An all-at-once release scatters the audience across the season within days. Discussion becomes hazardous because any given viewer may be several episodes behind, so it peaks fast and fades quickly.
That difference matters most for titles that depend on being talked about. A show with a strong existing audience needs the conversation less than a new one does.
The viewing curve is not the same shape
Full-season drops produce a tall, narrow spike. Most of the total viewing arrives in the opening days, which makes the title look enormously successful and then look finished.
Weekly releases produce a lower, longer curve that often rises as late joiners catch up. The peak is smaller but the title stays visible on the home screen for months rather than days.
Services increasingly split the difference, releasing several episodes together and then moving to a weekly rhythm, which buys an opening spike without surrendering the long tail.
Cancellation behaviour differs at the edges
Subscribers who join for a single title behave predictably. If the title is exhausted, the subscription has served its purpose and the account becomes a candidate for cancellation.
A weekly schedule keeps that decision permanently deferred. There is always an unwatched episode arriving, and the cost of cancelling includes losing the thread of something in progress.
Where the calculation reverses
Some formats resist weekly pacing. Comedy with light continuity, and anything a viewer treats as background, gains little from being rationed and can lose momentum entirely.
Licensed catalogue arrives complete by default. The cadence decision only exists for originals, which is why it correlates so closely with what a service is trying to prove about itself.