A pilot exists to turn assumptions into evidence. It should settle four things no document can: whether your subject holds attention on the record, whether the format survives its own runtime, whether a fifth and tenth episode genuinely exist, and whether you could make another one at the cadence you promised. It proves nothing about audience size.
Flow diagram of 5 steps: Write down your assumptions and what would disprove each, Start a time log before the first email, not before the first recording, Note how much you recorded against how much you used, Keep what failed, and write one line saying why, Show it to two people who have not heard you describe it.
Most people making something for the first time use the word pilot to mean episode one, made early and shown to somebody. That definition is common and it is not useful, because it gives you no way to decide how long the thing should be, how polished, or what it should be about. The narrower definition does all of that work: a pilot is an artefact made to answer a question that cannot be answered on paper. Decide which question, and the length, the subject and the level of finish stop being matters of taste.
The distinction changes what you actually make, and it changes it in an uncomfortable direction. Episode one wants to be the best possible introduction to the series, so it reaches for your easiest subject, your most cooperative contributor, the room you already know sounds good. A pilot wants to be the fastest possible test of the assumption most likely to be wrong. Those two aims pull against each other. The version that flatters the idea is very often the version that tells you least about it, because every variable that might have failed has been quietly removed in advance.
It is worth separating this from testing the idea itself, which is a different and much cheaper activity that belongs earlier. Saying the premise out loud to somebody and watching them fail to care costs five minutes. A pilot is what you make once the premise has survived that, and it tests execution rather than concept: not is this worth doing, but can this be done, by me, repeatedly, in the form I described. An idea can be excellent and still produce a programme nobody can sustain past the third week.
The practical consequence is that a pilot which confirms everything you already believed has probably been designed badly. It has cost a week and returned no information. Before you start, it is worth being able to say what result would make you change course — a different format, a different contributor, a shorter runtime, or dropping it. If no outcome would change anything, you are not testing, you are rehearsing, and rehearsing is a legitimate thing to do but it should be called by its own name.
These are the assumptions that survive every amount of planning and then fail in production. Each of them looks settled while the project is still a document. None of them is settled until something has been recorded.
Somebody can be compelling across a kitchen table and flat the moment a recorder is running, and the usual explanation — nerves — is only occasionally the real one. The more common problem is that a good conversation and a good recording want different behaviour. In conversation your contributor is entitled to answer in fragments, to assume shared context, to trail off once the point has landed, and you fill the gaps with your own knowledge. On the record those same habits produce answers that cannot be cut, because each one depends on a question or a piece of context that is no longer there. The failure mode to watch for is the two-sentence answer followed by a full stop and an expectant look. It is cheap to test: twenty minutes with the real equipment, on the real subject, listened back to the next day rather than in the moment.
Structure that carries five minutes will not necessarily carry twenty-five, and the failure is specific rather than vague. A short piece can be one movement — a single question, opened and closed. At around fifteen minutes an audience needs somewhere to have arrived, which means the piece needs internal joints: a shift of place, a second voice, a change in what is being asked. A great many first attempts are a five-minute idea stretched with repetition until it reaches the length that was promised. The only way to find out whether yours is one of them is to build at the real length. A three-minute cut of a half-hour format has tested a different programme, and usually tested the easiest three minutes of it.
This is the assumption that quietly kills more series than any craft problem, because a format can work perfectly once and have no supply behind it. The test costs nothing and almost nobody does it: before making the pilot, write down six further episodes with real names attached — this person, that place, this specific event. Not categories, not we would find more people like her. If you can name three and the rest is a hopeful gesture, you do not have a series, you have one strong piece, and one strong piece is a perfectly good thing to make deliberately. The distinction matters most for formats built on rare access, where the first contributor was rare precisely because there are not many of them.
A pilot is the only honest source of production cost, and only if you measure it while it happens rather than estimating afterwards. Record every hour, including the parts that do not feel like production: the emails to arrange access, the rescheduling, the travel, the twenty minutes spent finding a file. First attempts are always slower than steady state, so subtract genuine one-off setup — learning the software, buying and configuring equipment. Do not subtract an imagined future in which you get faster. If the pilot takes forty hours and the honest repeatable figure is twenty-five, a weekly programme is a twenty-five-hour-a-week commitment, and that number is worth knowing before you promise a weekly programme to anyone.
A full-length pilot is the most expensive way to learn almost anything, and for several of the questions above it is straightforward waste. The broadcast industry worked this out long ago and has a vocabulary for the cheaper options, which is worth borrowing even when there is no commissioner anywhere in the picture.
The table sets out the realistic forms and, in the last column, the thing each one will mislead you about. That column is the one to read against your own question, because every format on this list answers something and quietly hides something else.
Forms a pilot can take, what each settles, and where each misleads
| Form | What it settles | What it hides |
|---|---|---|
| Full pilot at the intended runtime | Structure, pace, whether the format has enough internal shape to hold its length, and the real production cost per episode | Nothing much, which is why it is the expensive option. The risk is spending it on a question a twenty-minute recording would have answered. |
| Presentation pilot — a shorter version, made to be seen rather than aired | Tone, look, the relationship between presenter and subject, and whether the thing feels like what you described | Pace and endurance. A trimmed version removes exactly the stretches where a format runs out of material. |
| Taster tape or sizzle — two to five minutes cut from real material | Whether the atmosphere you promised exists on the recording, and whether access is genuinely available | Almost everything structural. A taster is assembled from your best moments, so it proves those moments exist and not that there are enough of them. |
| One unedited interview or scene | Whether the contributor talks in usable answers, and whether the location is workable at all | Editorial shape entirely. This is a component test, not a programme, and treating it as one flatters material that an edit would expose. |
| A standalone piece that also works as episode one — the backdoor pilot | That the idea stands up as a finished thing in its own right, with no series required to justify it | Repeatability. A piece built to survive alone often uses up material or access that the second episode was going to need. |
| An audio-only version of a visual idea | Whether the underlying story carries without pictures, which is a hard and informative test | Everything the pictures were for. If the idea is fundamentally about how a place looks, this test will fail it unfairly. |
The information a pilot produces is perishable. Three weeks later you will remember that it went reasonably well and you will have lost the specifics that made it useful — which hour disappeared, which question fell flat, how much material you discarded. The following costs a few minutes a day during the production and converts the whole exercise from an impression into something you can act on.
Before any recording, list the three things that have to be true for this to work — she will speak freely about the closure, the room will be quiet enough, this exists in six other towns — and beside each one, the observation that would show it is not. Doing this afterwards is worthless, because by then you already know the answer and will write down the assumption you happened to test.
Count research, arranging access, travel, rescheduling, file management and the evening lost to a software problem. For first-time producers the recording is routinely the smallest block of time in the whole project, and a log that begins when the camera turns on will understate the real cost by a factor that makes a weekly cadence look plausible when it is not.
Record the ratio for each element — sixty-two minutes of interview producing seven usable minutes, for instance. It tells you two things at once: how much recording time to budget for the next one, and whether your questions are working. A ratio that is getting worse across a session usually means the questions drifted from specific to general rather than that the contributor faded.
The unusable takes are the most informative material you will produce, and the instinct is to delete them. One line each is enough: answer had no subject in it, room tone changed when the heating came on, I interrupted. A pilot that generates six such lines has done more for the series than one that went smoothly.
The condition is the important part. Anyone who has heard your explanation will watch the programme you intended rather than the one that exists. Give no setup at all, then ask one narrow question afterwards — where did your attention go, or what did you think it was about for the first minute. Asking whether they liked it produces politeness, which is not information.
At the end, in one place: hours for this episode, hours you believe a repeat would take, what you would change, and which of your assumptions turned out false. This is the entire output of the exercise. Everything else is a recording; this is the thing that tells you whether to commit, change the format, or make something else instead.
A pilot is evidence about production, and people routinely try to make it evidence about demand, which it cannot be. Showing a finished piece to friends, family and a few interested contacts produces a sample that is both tiny and selected for warmth. Enthusiasm from that group is not a signal that an audience exists, and indifference from it is not proof that one does not — several of them are reacting to you rather than to the work, in both directions. If you want to know whether there is an audience, that is a separate question requiring a separate method, and a pilot answers it neither way.
The second thing it cannot tell you is whether the programme will still be good at episode eight. Novelty is doing real work in a first episode, for you as much as for the audience, and it is not available again. Presenting improves with repetition, so an awkward pilot understates what the fifth one would be like; conversely the exhaustion of making the same thing weekly is invisible at episode one. Neither effect can be measured from a single artefact, which is one more reason to treat the supply question in advance rather than trusting that ideas will turn up.
It also cannot tell you that the piece is good in any durable sense. A pilot tells you whether it was makeable, whether it held together at length, and whether two uninformed people followed it. Those are narrow findings and they are the ones worth having. Judgements about quality made about your own work three days after finishing it are unreliable in both directions and are best postponed.
It is worth being plain about what sits behind a pilot here, because the word carries assumptions from an industry this platform is not part of. Kind Channel is new. Nothing has aired, the community is small, and there is no development fund, no commissioning budget and no slate to be added to. Nobody here will pay for a pilot, and making one buys no place in a schedule that does not yet exist. A pilot is not required in order to propose something, and a proposal is not judged on whether finished material exists.
What a pilot is genuinely useful for, in this setting, is the same thing it is useful for anywhere: it replaces claims with observations. A proposal saying the format would be a fifteen-minute interview is an intention. A proposal saying a fifteen-minute version was made, that the interview ran an hour to yield nine usable minutes, that the second and third contributors are already arranged, and that a repeat would take about eighteen hours, is a description of something that demonstrably works. That is more useful to anyone assessing it, and considerably more useful to you, since you are the person who has to make the next nine.
The test of whether a pilot did its job is simple and it is not about how the pilot turned out. The pilot worked if you now know something you did not know before, and if that knowledge changes what you would do next — even when the change is to stop. A pilot that leads to abandoning a format after one week has saved several months, and that is a good outcome rather than a failed one.
Long enough to answer the specific question you are testing, and no longer. If the uncertainty is about structure and pace, it has to run at the intended length, because those problems only appear once a piece has to fill its runtime. If the uncertainty is about tone or atmosphere, a two-to-five-minute taster cut from real material settles it for a fraction of the cost. If the uncertainty is whether one particular person gives usable answers, twenty minutes of unedited recording answers that completely. Building a full half-hour to find out whether a contributor speaks well is spending roughly twenty hours on a question that twenty minutes would have closed.
Usually a standalone, and for a practical reason rather than a creative one. Episode one carries setup that no other episode carries — establishing the premise, introducing recurring elements, explaining why the series exists — so it is both unrepresentative of normal production and disproportionately expensive to make. An episode that feels like it comes from the middle of a run tests the ordinary condition, which is the condition you will be in for every week after the first. The exception is a format whose premise genuinely needs re-explaining to a new audience each time, such as an anthology or an interview strand with no continuing thread. There, the setup work is the normal condition, and testing it is the point.
No. There is no requirement for finished or recorded material, and proposals are not assessed on whether any exists. It is also worth being clear about what a pilot would and would not get you: Kind Channel is new, nothing has aired yet, and there is no development fund or commissioning budget, so nobody here can pay for a pilot or guarantee that making one leads to a slot. The reason to make one anyway is for your own decision. A proposal that can state what a recording actually cost, how much material an hour of interview yielded and who the next three contributors are describes something tested rather than something imagined.
That depends entirely on which assumption failed, and separating the two categories is the most valuable thing you can do afterwards. Craft failures — muddy sound, a slack edit, questions that produced short answers, a first appearance in front of a camera — are expected on a first attempt and improve quickly with repetition, so they say very little about the idea. Structural failures are the serious ones: no further episodes you can actually name, access that turned out to be a favour rather than a standing arrangement, or a format with nothing to sustain its second half. A technically rough pilot of a structurally sound idea is normal and worth continuing. A polished pilot with nothing behind it is the harder result to accept and the more important one to act on.