Pitch an Observational Series Without Inventing What People Will Do

A pitch deck that promises "Episode 4: the coordinator walks out" has stopped pitching a series and started writing one. It's an easy slide to arrive at, because an episode list looks like a promise, and promises need events. But an observational series cannot promise events involving real people. What it can promise is a credible reason to keep the cameras running.

Those are two different documents, and the difference is the whole skill. An observational engine is a source of unfolding situations: a recurring activity that keeps producing decisions, pressures and unfinished business, which the team can describe because they've watched it, read about it, or been told about it. What it is not is a screenplay assigned to real people before they've agreed to anything.

Four questions hide inside that. What keeps happening? Can we film it? What might follow from it? What don't we know? Answering them in order is much easier than answering them all at once, which is what most decks try to do.

Describe what happens repeatedly before promising episodes

Start with the cycle, not the characters. Look for an obligation, a season, a rota, a weekly procedure — anything with a clock attached and a decision built into it.

Then write down how you know it recurs, and keep the answers apart:

• You watched it happen.
• Someone who lives inside it told you it happens.
• A record exists showing it happened.
• You assumed it happens.

These are not interchangeable, and the fourth is not evidence. Most weak pitches are the fourth wearing the clothes of the first.

A recurring activity does not need to look dramatic to support a series. It needs to generate decisions with visible consequences. Take an invented example that can run through this piece: Riverbend Community Radio, a small volunteer station. Its crews take a week each. They produce a half-hour community affairs program, and at the end of the week the outgoing crew writes a handover note for whoever comes next. Fourteen months of those notes exist, about sixty of them. Each one carries a running list of segments nobody finished.

That is not a conflict. It is a procedure. And it produces a recurring decision every single week: the next crew either finishes the carried-over segment, runs it late, rebuilds it, or drops it. Nobody has to shout for that decision to matter. If you find yourself reaching for conflict because the ordinary activity sounds quiet, you probably haven't found the decision yet.

Keep recurrence and filming access in separate columns

Two questions, two columns, no mixing.

Column one: does this keep happening? Column two: can we actually observe or film it happening? A strong answer to the first tells you nothing about the second. A hospital handover runs at seven every morning whether or not a camera exists. Riverbend's handovers happen weekly whether or not anyone films them. Recurrence is a fact about the world; access is an agreement between people.

Phrases to distrust:

"We have access." This answers neither column. It usually means the team visited once, or knows somebody.

A signed letter. A station manager's support is a real thing and worth having. It is rarely a filming arrangement. It doesn't tell you which rooms, which hours, which people, or whether anyone in frame can decline.

"The station is open to it." Open to what? An orientation session? A week of observation? Cameras on the floor during a live broadcast?

For Riverbend, column one is reasonably strong: sixty handover notes, plus volunteer and coordinator accounts that carried-over segments are common. Column two is nearly empty. The coordinator invited the team to sit in on one orientation session for new volunteers. That's the entire access record. Filming during a working week has not been discussed with anyone, and the volunteers who would appear in it have not been asked.

The sentence that captures this honestly is dull, and dull is correct here: "We have been invited to one orientation. Filming during shifts has not yet been discussed with crews." A pitch that says that is more useful than one that says "we have access," because a commissioner can act on it. They know what to ask next.

Show what can vary without scripting the participants

The engine isn't the repetition. It's the variation inside the repetition. Your job is to describe the range of ways the recurring decision could resolve, then say what you'd want to watch. The range is the series. The specific resolution is what you find out later.

The same entry in the log can mean three different things. A guest cancelled, or a crew ran out of time, or nobody booked the studio. All three leave an identical mark on the handover note and hand the next crew a completely different problem. You can list the explanations the pattern could produce. You cannot say which one happens in week three.

That's why the useful thing to write is an observation question rather than a behavior. Compare:

Weak: "A volunteer confronts the coordinator about airtime."

Strong: "What happens to a carried-over segment when the next crew already has a full slate of its own?"

The second one is a question the series can answer in several ways and still work. The first has already decided the answer, cast a role, and assigned a motive.

Here's the same material written two ways. First, defensibly:

Riverbend's crews hand unfinished segments forward every week, and the log shows which kinds keep coming back. We would follow the carry-overs themselves: what happens to a segment when three different crews inherit it, and whether the station ever kills one outright.

Second, with a story bolted on:

Episode 3: A new volunteer, Dana, refuses to let the market segment be dropped. The coordinator pushes back. By Friday, the station's whole approach to airtime has changed.

Nothing in the packet records a refusal, a pushback, or a change in how the station schedules anything. Dana has been given a motive and a transformation before she has agreed to be on camera. The second version is more exciting in exactly the way the evidence isn't, which is the trap. It might happen. If it does, you'll be delighted. You just can't build the case for the series on something you made up.

Distinguish established material from prospective coverage

Inside the pitch, use four labels and use them literally.

Documented event — a record or your own observation supports it. "The log marks carried-over segments across fourteen months of notes."

Reported pattern — people told you, and there's no record behind the telling. "Crews say they handle carry-overs differently."

Possible situation — something the pattern could produce that you haven't seen. "A segment that three crews in a row decline to run."

Unresolved question — what you don't know and aren't pretending to. "Whether any crew will agree to be filmed during a shift."

A five-way worksheet is worth keeping behind the scenes, because you need the extra resolution to keep yourself honest:

Now run the check on your own best idea. The three-crew episode rests on a reported pattern, not an observed one, because nobody has watched two working weeks side by side. So the deck sentence becomes "crews describe handling carry-overs differently," not "crews handle carry-overs differently." One word, and the pitch survives scrutiny.

The labeling habit has a published cousin. A 2025 nonfiction proposal checklist, the Nonfiction Core Application 3.1, asks applicants to identify unknown story elements rather than let them pass as settled (printed p.2, Film Summary). Two limits are worth stating plainly. It's application guidance for one organization's process, not a method for building an observational engine, and it doesn't establish that labeling makes a series work. What it does support is the habit of naming what you don't know instead of burying it in confident prose.

And watch your conditionals. "May," "could," and "we don't yet know" are not a blessing you sprinkle over a fixed ending. If the line reads "when the station decides to drop the segment, the crew reacts," the when has already assumed the decision. A conditional should be load-bearing: the episode changes depending on how it resolves. If the episode reads the same either way, the conditional is decoration.

Write the series promise at the level the evidence supports

Now assemble the three pieces: recurrence, meaningful variation, and the perspective that makes them worth watching.

For Riverbend, that's a paragraph, not a page:

Each week, volunteer crews inherit unfinished segments from the crew before them and decide what to do with them. We would follow that decision across different crews and different kinds of segments. Filming during shifts has not been arranged, and we don't yet know whether crews will agree to it.

Three sentences: what recurs, what we'd watch, what's unresolved. A commissioner can check every clause. That is what makes it persuasive, not the polish.

Sometimes the check tells you the evidence won't stretch. If sixty notes describe one handover crisis with a beginning and an end, you have a film, or a series with a known terminus. Say so. A pitch that quietly spreads one finite story across six episodes isn't lying exactly, but it's the same move as inventing a climax: the promise outruns the material.

Two things the pitch should not become. It isn't a production guarantee, so it shouldn't certify that the series will deliver scenes it hasn't seen. And it isn't a welfare or consent assessment, so it shouldn't imply that participant protections are settled when they're a separate conversation with separate people.

Three conditional episode opportunities come out of the Riverbend packet, each carrying its own requirement:

• One carried-over segment followed across three crews, if the log shows a category that recurs and crews agree to be filmed.
• A single crew's week, tracking everything it inherits and everything it hands on, if filming during a working week can be arranged.
• The log itself as a through-line — an accumulating record of what a small station can't quite finish — if the station will let the record appear on screen.

Notice that none of them requires a participant to do anything in particular. Each of them gets more interesting if someone does, which is precisely the difference between an engine and a script.

The invented station here is a teaching case, and the method it demonstrates is an approach worth testing rather than a proven one. There's no series-specific source behind it in this draft, and no filming or consent work has been done. What you can take from it: name the recurring decision, keep access in its own column, let the variations be possibilities, and say out loud what you don't know. A pitch that admits three unknowns isn't weaker than one that admits none. It's the only kind that survives contact with a real working week.

Frequently asked questions

Why is it a problem to promise specific events for real people in an observational pitch?

A pitch deck that promises 'Episode 4: the coordinator walks out' has stopped pitching a series and started writing one. An observational series cannot promise events involving real people. What it can promise is a credible reason to keep the cameras running. An observational engine is a source of unfolding situations: a recurring activity that keeps producing decisions, pressures, and unfinished business. It is not a screenplay assigned to real people before they have agreed to anything.

What four questions structure an observational engine?

What keeps happening? Can we film it? What might follow from it? What do we not know? Answer them in order. Start with the cycle, not the characters: an obligation, season, rota, or weekly procedure with a clock and a decision built in. Write down how you know it recurs: you watched it, someone told you, a record exists, or you assumed. The fourth is not evidence.

How should recurrence and filming access be treated?

Keep them in separate columns and do not mix them. Recurrence is a fact about the world; access is an agreement between people. A strong answer to recurrence tells you nothing about access. Distrust phrases like 'we have access,' a signed letter, or 'the station is open to it.' In the Riverbend example, sixty handover notes support recurrence, but the access record is one orientation session; filming during shifts has not been discussed, and volunteers have not been asked.

What is the difference between an observation question and scripting a participant?

An observation question can be answered in several ways and still work: for example, what happens to a carried-over segment when the next crew already has a full slate of its own? A scripted participant line has already decided the answer, cast a role, and assigned a motive: a volunteer confronts the coordinator about airtime. In the Dana example, nothing in the packet records a refusal, a pushback, or a change in scheduling. Dana has a motive and a transformation before she has agreed to be on camera.

How should a pitch label evidence and unknowns, and what should the series promise?

Use four labels literally: documented event, reported pattern, possible situation, unresolved question. A three-crew episode rests on a reported pattern, not an observed one, so the deck says crews describe handling carry-overs differently, not that they handle them differently. Promise at the level the evidence supports: what recurs, what you would watch, and what is unresolved. If the evidence will not stretch, say it is a film or a series with a known terminus. The pitch is not a production guarantee and not a welfare or consent assessment.

Further reading

• Show How the Crew’s Presence Shapes an Observational Series — After the access column, read this to describe the crew's own recurring presence: what the production arranges, asks and starts, and why that isn't a guarantee of candour.
• Pitch a Documentary Series Built From a Public Record — When your evidence is a log, this shows how to bound the collection and state what each record can and cannot support before building episodes on it.