Pitch an Investigative Series When Each Episode Tests a Different Explanation

A factual team arrives with a question, six interviews lined up, four locations, and an episode list that reads like a guest register: the former manager, the family, the site, the archive. Every episode has a subject. None of them can fail.

That is the problem. If nothing in episode three can come back the wrong way, episode four was never really contingent on it. You have a running order, not an inquiry. The person in episode five repeats the question the person in episode two has already answered with conviction and no evidence, and the series ends where it started, forty minutes later, with better lighting.

The fix is not to add suspense or withhold the good material until the end. It is to build each episode around a test the evidence can pass or fail, and to write the pitch so the reader can see what happens on either outcome. This article works through one invented case from first question to finished inquiry map.

The case is fabricated teaching material, and so is everything in it: the cinema, the town, the newspaper, all three documents. Nothing here describes a real venue, a real dispute, or records anyone has actually examined.

State the question before listing explanations

Sort what you already have into three states and keep the words in the pitch.

Established means a fact about the record: what exists, what it physically is, who holds it, what it actually says. Disputed means two supplied accounts cannot both be right as stated. Proposed means an explanation you are considering, with nothing yet behind it but plausibility.

Teams usually blur these, because the blur flatters the outline. "The cinema opened in 1948" sounds like a finding. In our invented case it is a sentence printed on page two of an anniversary leaflet, and that is all it is.

Here is the case. The Orchard Cinema, in the invented town of Wrenfield, appears in local memory with three opening dates. A printed preview invitation, the size of a postcard, reads "Friday 9 September" with the year 1949 added in pencil by a hand nobody can identify, and a manuscript line addressing it to the Editor of the Wrenfield Advertiser. A hardback ticket ledger's first used page is headed "October 1949"; its lines give the date a booking was taken, the performance date, seat numbers, a price, and a name or the word "cash." Every performance booked on that page falls on or after Monday 31 October 1949. Those pages carry no totals. From November the book changes shape: each page is headed with one performance date and a column that looks like a nightly take. And a four-page leaflet, cover-dated 1999 and headed "Fifty Years of the Orchard," says on page two that the cinema opened in 1948. That leaflet argues with itself: its title counts fifty years from 1949, whether from the September or the October date, while page two names 1948. It is a disputed document before anyone has left the room.

What is established? Three documents exist and say what they say. That the preview was held, that the October performances ran, and where 1948 came from are all open. So is the question of whether the three dates are even competing.

That last point matters more than it sounds. A private preview and the start of a paying programme describe two different events. A cinema can have both, a few weeks apart, and both can be called an opening with a straight face. Do not stage a contest between accounts that are answering different questions. The useful question here is not which date is correct but what was each date recording at the moment it was made. That is bounded, it is answerable from evidence, and an answer would change what the series says.

Now write your list of explanations, and put each one through a single test: name one thing that would support it and one thing that would weaken it. An explanation for which you can name neither is not an explanation. It is a mood with a title card.

One more rule before the list goes in the pitch. If an explanation implies that a named person did something wrong, the pitch has to name the conduct, the standard it breaks, and the evidence that would show it. If you cannot supply all three, the explanation does not go in the outline. Nobody becomes a suspect because the deck needed an antagonist. Our cinema example has no antagonist at all, and the inquiry is not weaker for it.

Choose evidence that could distinguish the possibilities

There is a difference between a reporting activity and an episode inquiry, and pitches collapse it constantly. "Interview the manager" is an activity. "Test whether the preview and the public opening were separate events" is an inquiry. The first fills time; the second can fail.

The test for a candidate piece of evidence is comparison. Ask what this record, encounter, or observation would look like if explanation A were true, and what it would look like if explanation B were true. If it looks the same either way, it cannot distinguish anything, and the episode built on it is a detour.

Take the invitation card. It establishes that a printed run existed and that at least one copy was addressed to a named editor. It does not establish that the card was posted, that the editor came, or that the room was lit. It also does not fully establish its own year. The printed weekday helps: a Friday 9 September comes around every few years, so "Friday" narrows the field without settling it. That is what the pencilled 1949 is being asked to do, and the pencil is doing interpretive work, not evidentiary work.

Evidence that could go further: the Advertiser's issue for the following week, which would either carry a report or not; a guest list or signature book; a dated photograph; the distributor's paperwork for a preview print. Each of those is a different question with a different failure mode, and each is worth planning separately.

Now the ledger, which is the centre of the case. Its October page was compiled by someone taking reservations for nights that had not happened yet. That is what a booking page is. From November, the book records a different kind of thing: figures entered after a house closes. The change in the book's shape is itself a finding about practice, and it points at the specific question worth an episode.

Were the tickets on the October page merely booked, or were the performances they reserved actually screened?

That question has real discriminating evidence behind it. Houses accounted for on those nights, in the same hand and ink as a night's other entries, are records made by someone present when the performance was supposed to run. A distributor's invoice charging for the print in that week is an independent record made by a different organisation for a different purpose. A press advertisement in the week beginning 31 October saying "now showing" is weaker than either but still relevant; one saying "opening Monday" establishes intention, not occurrence.

Two records are worth more than two documents when they were made by different processes. The ledger and the invoice come from different desks and would each have to be wrong in a different way. A 1999 leaflet quoting a 1949 programme note, by contrast, is not corroboration. It is the same claim travelling in a nicer jacket. The leaflet's own cover is a smaller instance of the same problem: it counts fifty years from 1949 while its page two says 1948, so even the document that carries the 1948 claim declines to agree with itself.

That is the discipline to apply across the whole outline. Before an episode goes in, write one sentence naming the comparison it performs. If you cannot, you have written a location with a presenter standing in it.

Write the contrary-finding branch

Every episode gets three branches, and the pitch should show work in all three: the evidence turns up as expected, the evidence is absent or ambiguous, and the evidence points the other way. Absence and refutation are different branches, and they lead to different next questions.

Most pitches handle this by adding a line under the episode: if the records do not support this, the series will adapt. That line is a disclaimer, not a branch. It does nothing for the reader, because it commits the series to nothing and describes no next question.

Give each branch its own paragraph and its own question. Here is Episode 2 written with all three.

Were the October bookings ever screened? Position: the ledger records reservations for performances from 31 October 1949 and accounts for none of them. Test: whether those pages, or a separate cash record, account for the houses on those nights, and whether the distributor billed for the print that week. If accounted for: the paying programme began at the end of October, and Episode 3 turns to why later publicity used 1948. If the record stays silent: Episode 3 becomes a question about the record itself — how this book was kept, how it was used, and what its shape can and cannot show. The hunt for a hidden opening is not over; it now depends on records nobody has located yet. If it points the other way: the reservations were cancelled, or the distributor's paperwork places the print elsewhere that week. The October nights were not screened, the paying programme began later than the ledger's first page suggests, and Episode 3 asks what the Orchard was doing in the interval.

Note what the silent branch does not say. It does not say the performances did not happen. A gap in an incomplete collection is a gap. Ledgers lose pages; pages lose ink; a second cash book may sit in a different box in a different building. Absence in a partial record weakens a claim without disproving it, and a series that treats every gap as a refutation is making the same error as the series that treats every document as confirmation.

The third branch is the one most pitches omit, because a contrary finding feels like the series going wrong. It is the opposite. If the reservations were cancelled, or the distributor's paperwork puts the print elsewhere that week, the evidence has actively ruled the October opening out rather than merely failed to record it — a firmer result than either of the other two branches yields.

The contrary branch should be the most interesting writing in the pitch, because it is where the series proves it is following the evidence rather than casting it.

Carry the same structure through the other episodes. For the preview, the branch for "no press report" is not a shrug: it moves the inquiry toward how the Orchard ran its invitations and who its publicity was aimed at. For 1948, the branch for "the year appears nowhere before 1999" is a change of subject — from a hidden earlier opening to how the anniversary leaflet was made, who wrote it, and what they worked from. The leaflet's own title already points one way: a cover counting fifty years from 1949 suggests its compilers treated 1949 as the start, or at least as the date worth commemorating, which makes the page-two 1948 the sentence that needs explaining. That is a real episode with real interviews, and it is available whether or not any earlier use of 1948 exists.

Where the branch leads to a person's account of their own conduct, keep the sequence honest about what has and has not happened: a plan to seek an interview is a plan, not a confirmation that anyone will speak.

Assign people and places to the inquiry

Once episodes are inquiries, the interview list stops being a cast and becomes a set of capabilities. Ask what each person or place makes possible.

Three kinds are worth distinguishing. A witness has direct knowledge of an event. A custodian holds or can interpret a record. A comparator can say what normal practice looked like — how a booking ledger in that trade was usually kept, whether a preview would typically be press-only, how a distributor's invoice would ordinarily be raised. A comparator supplies the standard against which you read a document, and pitches rarely think to include one. In a case this old, the comparator is often a professional or an enthusiast with grey literature, not an eyewitness.

Be honest about time. Witnesses to a 1949 event are, at best, very old, and most of the useful memory of it now sits two or three retellings away. A person describing what a parent told them is a source, but they are not a firsthand account of the evening, and the pitch should say which one it is offering. In our fictional case the series will lean on records, and its interviews will mostly be with custodians and comparators explaining how to read them. That is a fair description of a working series, and writing it plainly is better than implying a room full of first-person recollections that no longer exist.

Then let the questions allocate the people, not the reverse. The local studies librarian serves Episode 1 and Episode 3: she can say whether the Advertiser's 1949 run is complete, and she can locate the earliest appearance of 1948. That one source answers to two questions, and the outline should show it rather than inventing a second person to keep the episodes tidy. One question may take two encounters. A single source may touch three episodes.

Locations work the same way. A site visit earns its place when the place can establish something — the position of a projection box, the size of an auditorium against a seat count in the ledger, whether a listed frontage was built before or after a date in dispute. Without that, it is scenery.

One practical consequence: name episodes for what they test, not for who appears in them. "The Manager" tells the viewer to expect a verdict about a person. "Were the October bookings ever screened?" tells them what the episode is actually for, and does not promise a villain.

Present the sequence as conditional progress

The pitch should show three things per episode: where the evidence currently stands, what this episode tests, and how each outcome would redirect the next question. That is the whole method, and it fits in a paragraph.

Show it as a page, not a promise. The reader is not being asked to believe a conclusion; they are being asked to believe the team can follow a finding. Write the current position from the three-state list you built in the first section. Write the test as a comparison, not an activity. Write every outcome, with the next question each produces.

Then draw the map. Expect to draw it twice.

The inquiry map, as it would be drawn before reporting

WHAT WAS EACH DATE RECORDING?

├─ EP 1  Was the 9 September preview held?
│    ├─ press report, guest list, print paperwork ─▶ how were invitations used?
│    ├─ nothing found ────────────────────────────▶ same question, wider search,
│    │                                               incl. complete press run
│    └─ a record contradicts the date ────────────▶ what was the printed card
│                                                    for, and who was it for?
│
├─ EP 2  Were the 31 October bookings screened?
│    ├─ houses accounted for that week ───────────▶ EP 3: paying programme began
│    │                                              31 October — so where did
│    │                                              1948 come from?
│    ├─ bookings only, nothing accounted ─────────▶ EP 3: how was this book kept,
│    │                                              and what can its shape show?
│    └─ reservations cancelled, print elsewhere ──▶ EP 3: what was the Orchard
│                                                   doing before it opened?
│
└─ EP 3  Where does 1948 first appear?
     ├─ in a document before 1999 ─────────────────▶ what was 1948 used for then?
     ├─ first in the 1999 leaflet ─────────────────▶ how was the leaflet made,
     │                                              and from what?
     └─ attached to another building or claim ─────▶ whose opening was 1948, and
                                                     how did it reach the Orchard?

Mark this in the pitch as a working drawing. It is a picture of how the evidence would redirect the series. It is not a schedule, and it is not a claim about what the series will find. If the cover letter presents it as a prediction, the reader will grade the pitch on whether the predictions came true, which is not what you want them thinking about.

Now redraw it after the result most pitches are afraid of — the ledger shows reservations and the nights are never accounted for.

The same map after that result

EP 2  RESULT: bookings only; no accounting for the October nights.

└─ EP 3  Was: what did the programme from 31 October consist of?
         Now: how was this ledger used and kept, and what does its
              shape establish?
     ├─ a second cash book is located ─────────────▶ the houses, and the
     │                                               October opening, come back
     │                                               into play
     ├─ no second book exists ─────────────────────▶ how did the Orchard record
     │                                               its early years, and what
     │                                               did it stop recording?
     └─ a second book shows the nights cancelled ──▶ the October opening is ruled
                                                     out, and EP 3 asks what came
                                                     before it

The arrows have moved. Episode 3 is still third, but it is a different episode, asking a different question of different people, and the pitch can show that in advance. That is what makes the series investigative rather than retrospective. The momentum a pitch needs is not a cliffhanger that implies more than the reporting supports. It is a sharper question, arriving at a point the reader can see coming and did not expect to be so specific.

Write the map, then write it again for each outcome you are not hoping for. If a redrawn version is empty — if nothing changes — that branch was not testing anything.

Frequently asked questions

What separates an investigative episode from a running order?

A running order has episodes with subjects but no test that can come back wrong. An investigative episode is built around a test the evidence can pass or fail, and the pitch shows what the series does on each outcome.

How should the Orchard Cinema's three opening dates be framed?

Not as a contest between three competing dates. A private preview and the start of a paying programme can be different events, so the useful question is what each date was recording at the moment it was made. The 1999 leaflet is already disputed because its title counts fifty years from 1949 while page two says 1948.

What do established, disputed, and proposed mean in this method?

Established means a fact about the record, such as what exists and what it says. Disputed means two supplied accounts cannot both be right as stated. Proposed means an explanation under consideration with nothing yet behind it but plausibility.

What evidence could show whether the October 1949 bookings were actually screened?

Discriminating records include houses accounted for on those nights in the same hand and ink as the night's other entries, or a distributor's invoice charging for the print that week. A press advertisement is weaker; one saying 'now showing' is relevant, one saying 'opening Monday' establishes intention, not occurrence. The booking page alone records reservations made before the nights happened.

What if the record stays silent on the October performances?

Do not treat silence as proof they did not happen. Absence in an incomplete collection weakens a claim without disproving it. The next episode becomes a question about how the ledger was kept and used, and what its shape can show, while the search for a second cash book continues.

Further reading

• The Documentary’s Question Is Answered Before Filming Ends. What Is the Film Now? — Go here once reporting starts and a result arrives early, to decide whether the pitch now describes a finished inquiry or one the finding has opened.