I have abandoned several series after one episode and later discovered, on somebody's insistence, that the show became something quite different by episode four.
That is not an accident of quality. Pilots are structurally unlike the episodes that follow them.
What a pilot is actually for
Historically, it existed to sell the series to a network. Its audience was a small number of executives deciding whether to order more.
Which means it is optimised for demonstrating a premise, not for being a good hour of television.
It has to establish every character, explain the situation, show the format, and prove the concept can sustain a series, all in the time an ordinary episode uses to tell one story.
The result is dense, exposition-heavy, and frequently the least natural episode in a run.
Why they are made differently
The production conditions differ in ways that show on screen.
A pilot is frequently shot months or years before the rest, sometimes with a different director, occasionally in a different location, with a budget structure that does not apply afterwards.
Cast changes between pilot and series are common. Roles get recast, sometimes after the pilot has been shot, and the pilot is either reshot or quietly adjusted.
Sets are rebuilt, tone is adjusted, and the writing team that produces the series is frequently assembled after the pilot is made.
Which means the pilot can be the work of a different creative group than the show it introduces.
The two failure modes
Pilots go wrong in two recognisable directions.
Overloaded, where so much has to be established that nothing is allowed to breathe. Characters announce their own characteristics. The situation is explained rather than shown.
Or unrepresentative, where the pilot is a strong self-contained hour that misleads about what the series will be, and viewers arrive at episode two expecting something else.
The second is arguably worse, because it converts an audience that then leaves.
What changed with streaming
The commissioning model shifted and the effects are visible.
Straight-to-series orders, where a whole season is commissioned without a pilot being made, are now common. That removes the sell-to-executives function entirely.
The consequence is first episodes that behave more like first chapters — slower, less self-contained, assuming you will continue.
Which is better in some respects and produces its own problem, since a first episode that establishes mood without delivering anything gives a viewer little reason to continue.
The measurable version of this is the widely discussed drop-off between first and second episodes, which platforms watch closely and which has pushed some commissioners back toward front-loading incident.
The three episode convention
Widely cited in fan communities and reasonably well founded.
By the third episode a series has generally finished establishing, the production has settled, and what you are watching is what the show will be.
My own experience matches this closely enough that I now apply it deliberately to anything recommended by somebody whose taste I trust.
For anything I have no particular reason to watch, one episode remains a reasonable filter, on the grounds that time is finite.
The counter-argument
Worth stating fairly.
A show that cannot make a case for itself in an hour is asking a lot, and there are more series available than anybody can watch.
The three episode rule is a reasonable investment when the supply is limited and an unreasonable one when it is effectively infinite.
Which is probably why it is quoted more by people who write about television than by people who simply watch it, and why commissioners have started insisting that first episodes work on their own regardless of what the writers would prefer.
The recut pilot
A practice worth knowing about because it explains some odd first episodes.
Pilots are frequently re-edited after a series is ordered, sometimes substantially, with scenes reshot, characters removed and narration added to smooth over changes.
Occasionally the pilot is not broadcast first at all, with a later episode used as the opener and the pilot shown out of order or shelved entirely.
Which is why some first episodes contain details that never recur, or characters who vanish, or a slightly different tone that resolves by episode three.
How to decide whether to continue
The practical question, since time is the constraint.
The signal I have come to trust is not whether the first episode was good but whether anything in it suggested the writers knew what they were making.
A confident premise executed clumsily generally improves. A confused premise executed slickly generally does not.
The other reliable indicator is dialogue. Production values, pacing and structure all improve as a show settles. Writers who cannot write dialogue in episode one are unlikely to learn by episode six, and that is the thing worth judging early.
Backdoor pilots
A variant worth knowing about. Some series begin as an episode of an existing show, introducing characters who then get their own series if the response is good.
These generally make better first episodes, because the format is already established and the audience arrives with context.
They are also confusing to encounter later, since the first episode of the resulting series is frequently not the introduction at all.