Pitch a Modular Commercial Without Promising That Every Combination Works

A modular campaign is not a pile of interchangeable assets. It is a set of complete films that are permitted to touch.

Most modular pitches skip that distinction because the arithmetic is so seductive. Two openings, two demonstrations, two endings: eight versions. Put the number on a slide, and it reads as reach. But eight is a permutation count, not a promise. It says nothing about whether a viewer who meets Opening B has already been told what the product is, or whether Ending B refers to a refill action nobody in Demonstration A performed. What you can honestly sell is a version set: the specific complete films you have actually assembled, watched end to end, and found coherent. Everything else is raw material.

The advanced decision underneath is a trade. You can build modules self-contained enough to join almost anything, and pay by repeating exposition, flattening specificity, and making every scene slightly more generic than the last. Or you can keep scenes that are strong on their own and accept a narrower set of permitted pairings, alternate pickups, and dedicated bridges. There is no universally correct side. There is only the side that matches what this campaign has to say and what it can actually produce.

Define the unit that can vary and the promise that cannot

Start by naming what each module does, not how long it runs. A forty-second block is not a module in any useful sense; a block that establishes a situation, demonstrates a use, explains an offer, or closes an encounter is. The job is the unit, because the job determines what the module assumes on entry and guarantees on exit.

Then write down the invariant: the proposition and product facts that every complete version must preserve. Not a tagline — the thing the tagline is trying to convey. Together with any offer eligibility rules and required information, this invariant is the fixed spine. Modules vary; the spine does not. A cut that preserves every product fact but reverses the emotional proposition is not a variant. It is a different campaign sharing a logo.

Here is where a good pitch separates itself from a deck full of squares. Length is not evidence of interchangeability. A module qualifies as interchangeable only if it enters and leaves with compatible knowledge, action, and product state. Those three words do real work:

Knowledge — what the viewer already understands when the module starts, and what they understand when it ends. Action — the concrete thing a character has done or is doing, and whether some later module depends on it. Product state — the condition of the product on screen and what it implies about what has already been shown.

Now the invented example. Call it a container campaign: a reusable lidded food container with a snap-lock seal and a replacement gasket. Three variable slots, two options each.

Openings

• O1 — Introduction. Establishes the container on a counter, names the snap-lock seal, shows the hinge, states the one-line proposition. Assumes the viewer knows nothing.
• O2 — In use. Assumes the container and its seal have already been introduced. Opens mid-kitchen, a character packing lunch, and spends its time on the seal's behavior rather than explaining that the product exists.

Demonstrations

• D1 — Sealing. Shows the seal locking shut over a filled container. Does not mention the gasket or refills.
• D2 — Refill. Shows the worn gasket being popped out and a replacement clicked in. Treats the seal as already understood, because the refill act only makes sense once the lid is known.

Endings

• E1 — Generic close. Returns to the proposition; refers to no specific demonstrated action. Works after any demonstration.
• E2 — Refill close. Refers specifically to the gasket swap the viewer has just watched — a character in the kitchen still holds the replacement, or the tray of gaskets remains in frame — and ties the offer to the refill habit.

Six modules. Eight mathematically possible complete films. The rest of this piece is about which of the eight you may put your name on.

Inspect the joins that arithmetic cannot test

Read each candidate join the way a viewer will: as one continuous experience, with no knowledge of which square a scene came from. The failures are rarely dramatic. They live in unresolved pronouns, people who were never introduced, speakers that change without warning, product actions that contradict an earlier state, and endings that assume an event nobody saw.

Take the eight. Assuming the invariant — container, snap-lock seal, gasket — is present in all of them, and that O2 and D2 carry the knowledge assumptions above, with O1 alone licensed to introduce the seal from scratch:

Three composite versions pass on their own, three are conditional on prior placement or on the proposition being added to O2, and two are excluded. Both failures are the same seam: any D1 with E2. The ending refers to a demonstrated refill action, and D1 has no refill. A viewer who watches that version has been told the container seals well and then, moments later, that the refill habit matters — a habit they have never seen performed.

The O2 + D1 + E1 row is worth dwelling on, because the join itself passes while the version is the one most likely to be quietly wrong. O2 assumes the seal was introduced. D1 shows it locking. So far, so good: the demonstration can carry an already-primed viewer, and nothing at the seam looks broken. But O2 never states what the product is or what it promises, and E1 assumes that proposition has been stated somewhere. If O1 was the intended source of the proposition, then O2 + D1 + E1 is a coherent film with a hole in it that no join inspection catches, because the missing fact is upstream. The join check clears it; the version check does not. That gap is why the table marks it Conditional rather than Passes. The repair is to add the proposition to O2 itself — a line the module does not currently carry — or to accept that O2 versions belong to audiences already reached by O1 material in another placement. No ending in this set restates the proposition: E1 assumes it was stated, and E2 ties the offer to the refill habit, so there is no restating close to pair with O2 unless one is written. It is a media strategy question wearing story clothes.

Now the important distinction. Some seams are craft problems; some are meaning problems. A jump cut at the join, an audio bed that doesn't match, a speaker whose mouth we can't see — those are transitions, fixable in the edit. They cost time and skill, not integrity. D1 + E2 is not that. The meanings cannot coexist: the ending's proposition depends on an action the film never performed. You cannot edit your way across a missing event. No amount of match-frame discipline makes a viewer remember a refill they never saw.

And when offer eligibility or required information enters the picture, the join has legal and disclosure weight, not just narrative weight. If E2 is where the refill offer terms appear, then D1 + E2 is not merely incoherent — it is an incoherent version of a regulated communication. Whatever the actual rules in a real market, the structural principle holds: the module carrying required information cannot be joined to a module that contradicts or undercuts the conditions it states. Flag those joins in the map and treat them as excluded on two independent grounds.

Compare broad compatibility with richer conditional versions

The eight-cell table exposes the real fork, and it is not the one most pitches present.

Route A: redesign for width. Rebuild D1 so it carries the refill itself — a beat, at the close of the sealing demonstration, where the character pops the worn gasket out and clicks the replacement in. Now E2's reference has an antecedent in every demonstration, and the D1 + E2 seam closes. Showing the gasket as merely visible or worn would not do it: E2 points at the swap, not at the gasket's condition. The count rises, but it does not reach eight: O2 still assumes an introduction it never gives, so O2 + D1 + E1 remains conditional on the proposition being added to O2 or carried in by placement, and O2 + D1 + E2 moves from excluded to conditional — the D1 + E2 seam is gone, the O2 dependency is not. The cost is specific: D1 gets longer, spends time on a component it wasn't about, and dilutes its sharp single demonstration. You have also promised the production that every D1 carries the refill beat, which constrains set dressing, props continuity, and pickup coverage. If D1 appears in four versions, that constraint ships four times.

Route B: keep the specific scenes, narrow the set. D1 stays clean. Accept three self-contained versions and three O2 versions only under the stated placement or proposition dependency, instead of promising eight. Publish both E2's dependency and O2's. This is the more honest version set, and it is frequently the stronger one creatively, because the modules that survive are the ones with a clear job rather than the ones engineered to be all things to all joins.

There is a third route, half-built into this table already, which is Route C: buy a bridge. To make D1 + E2 viable, you may be able to add material — a short refill pickup at the join, or a re-cut of E2's opening line so it no longer points at a specific prior action. Where Route A builds the refill into every D1, Route C buys the same material once, as a pickup. A bridge is not a free fix. It is an added production requirement: a pickup shoot day or an extra coverage setup, new footage to log and clear, a new join to inspect, and a new film to count. In this example, the cheapest genuine bridge is a two-second pickup: the character's hand lifting the replacement gasket from the tray as the seal closes. That makes E2's reference true wherever the seal has been primed, and closes the D1 + E2 seam specifically. It does not by itself restore the O2 proposition, so the count returns only as far as the knowledge dependencies allow. But it is footage that did not exist in the plan, and it belongs in the budget conversation, not in a footnote on the permutation count.

What the three routes reveal is that the diagram is the wrong artifact for the decision. If someone shows you six squares and eight lines, ask what each scene's inbound knowledge is and whether the outward one is the same. If the answer is "we'll fix it in the edit," the case is not made.

And resist the softer versions of that answer: fewer reusable blocks on a diagram does not imply a cheaper shoot, a lighter edit, or an easier review cycle. Route B's six versions might be more expensive to produce than Route A's eight if the narrower set demands alternate pickups. Route A's universality might cost more review cycles because every version shares D1 and every D1 review pulls in the same D1 notes. Count work, not blocks.

Make exclusions part of the creative plan

The deliverable is not a diagram of nodes. It is a compatibility map with reasons, and a set of complete selected films, each watched whole.

A compact version:

Allowed

• O1-D1-E1 — proposition stated, seal shown, generic close.
• O1-D2-E1, O1-D2-E2 — gasket introduced by O1, refill shown, close matches.

Conditional

• O2-D2-E1, O2-D2-E2 — seal primed offscreen by prior placement, refill and close consistent. Conditional: the placement that primes the seal must also have carried the proposition, since O2 does not state it.
• O2-D1-E1 — seal primed, demonstration delivers, close generic. Conditional: requires the proposition to be added to the O2 scene or carried in by prior placement. Mark the dependency; do not bury it.

Excluded

• D1 + E2 — E2's refill reference has no antecedent. Reason: missing action in the film. Bridge to enable: gasket pickup at the D1/E2 join, budgeted as added footage.

That is a version set you can put in front of a client without wincing. It names what varies, what is preserved, what is forbidden, and what it would cost to change your mind. The conditional rows are the professional tell: they show you know the difference between a version that is internally coherent and a version whose coherence depends on where it runs.

Then inspect the selected complete films, not the modules. Take the three self-contained versions and whichever conditional O2 versions the placement or proposition plan actually enables, and play them end to end. Two things will happen. You'll catch joins that looked fine on paper — a pronoun the viewer can't resolve, a hand that changes position, a product state that flickers. And you'll discover the ends of the process are where the value leaks: a version that is technically coherent but says nothing, or says it twice. That is a story decision, and the map can't make it for you.

Any real campaign needs the actual version rules, the authorized assets, and someone's production account before anyone can claim a version set is feasible, or that it saves work, or that audiences understand it. This table is reasoning, not testing. None of these versions has been edited. The pickup does not exist. No one has watched them.

The bounded set is the pitch

A credible modular proposal ends by naming its limits. Not as an apology — as the shape of the thing. Which films can be assembled, which cannot, and what material would be needed to change the answer. The forbidden join is not the system failing. It is the system, drawn honestly. A pitch that promises eight out of eight has either done the joins and earned them, or it hasn't looked.

Frequently asked questions

What is the difference between eight possible versions and a version set?

Eight is a permutation count: the mathematical combinations of openings, demonstrations, and endings. A version set is the specific complete films you have assembled, watched end to end, and found coherent. Only the version set can be honestly sold; the remaining combinations are raw material.

What three things must match for modules to be interchangeable?

Knowledge, action, and product state. A module qualifies only if it enters and leaves with compatible knowledge of what the viewer understands, compatible action that later modules may depend on, and compatible product state on screen.

Why does any D1 with E2 fail in the container example?

E2 refers specifically to the gasket swap the viewer has just watched, but D1 shows sealing and never mentions the gasket or refills. The ending's proposition depends on an action the film never performed; that is a missing event, not a transition that can be fixed in the edit. If E2 carries offer terms or required information, the join also has disclosure weight.

Why is O2-D1-E1 marked Conditional rather than Passes?

The join itself works: O2 assumes the seal was introduced, and D1 shows it locking. But O2 never states what the product is or what it promises, and E1 assumes the proposition has been stated somewhere. The missing fact is upstream of the join, so the version check catches what join inspection does not. The repair is to add the proposition to O2 or rely on prior placement that carried it.

What are the three routes for handling incompatible joins?

Route A redesigns for width, such as building the refill beat into D1, at the cost of length, dilution, and production constraints. Route B keeps the specific scenes and narrows the set, publishing dependencies instead of promising eight. Route C buys a bridge, such as a two-second pickup at the D1/E2 join, which is added footage and budget, not a free fix. Count work, not blocks. The table is reasoning, not testing: no version has been edited, the pickup does not exist, and no one has watched them.

Further reading

• Pitch a Campaign That Teaches a Product One Step at a Time — If the set depends on which film a viewer sees first, read this for mapping prerequisites and separating authored order from actual viewing order.
• Write an Irreversible Product Demonstration Without Hiding the Change — For demonstration modules that turn on a physical change, this explains which footage records the action and why a cut can imply a connection no frame shows.
• Read a Commercial Bid Scope Before Describing What the Budget Covers — Once a bridge or pickup enters the plan, this helps read what a bid scope actually includes, excludes or leaves unmentioned before promising the added footage.