When an Interactive Commercial Needs a Different Story From the Linear Film

By pitch.dogAdvanced14 min read20 September 2026

Advertising

Your spot is cut, graded, and approved. Now someone wants a version where the viewer does something. The first question is not what the button looks like. It's what the viewer knows afterward that they wouldn't have known otherwise — and whether the film you already cut can survive that answer.

I'll work this through one invented example, because the useful distinctions only show up when the same story gets dragged through both versions. Nothing here describes a produced commercial; the film, the brand, and the branches are made up so the arithmetic can be checked.

The film. A bag brand — call it Wend — has a forty-second spot called "Sunday." Mara's kitchen, rain on the window. She cuts bread and makes two sandwiches, wraps both. She packs a tote: thermos, folded blanket, one enamel cup. Then a beat, and a second cup goes in. No phone, no dialogue, no explanation of the second cup. Doorbell at twenty-six seconds. Mara opens the door and Ines — the neighbour, soaked, holding a flat of strawberries under her jacket — is standing on the step. Mara reaches into the bag and hands her the second cup and the second sandwich. They sit on the porch under cover, the bag between them. End card.

Forty seconds, one path, and the second cup is the whole engine. The audience sees it, can't account for it, and spends the next sixteen seconds waiting for the answer.

An interactive route can't get that engine by adding a button. It has to redistribute what the audience is allowed to know. That redistribution is the actual work, and it changes the edit more than it changes the plot.

Describe the action and its narrative consequence

The proposed interaction: at eighteen seconds, as Mara closes the bag, the viewer is offered two accounts — Stay with Mara or Go with Ines — and the piece continues on the chosen side of the door.

Look at what each gives the viewer, because it is knowledge, not outcome.

Stay with Mara: she takes her phone off the counter. The screen shows a message she sent this morning with a time and an address, and nothing under it. She reads it, puts the phone face down, and goes back to the shelf for the second cup. The viewer now knows the invitation went out and went unanswered, and that Mara packed anyway.

Go with Ines: we leave the kitchen and pick her up on the pavement two streets over, hood up, strawberries under her jacket. She stops under an awning, reads the same message on her phone, and doesn't type a reply. She looks at the rain and walks on. The viewer now knows the message was read, and that the answer is being walked over rather than sent.

Same fact — the invitation wasn't answered — seen from two positions. Nothing about the ending changes. What changes is what you carry into it. One viewer arrives at the door knowing Mara waited without knowing; the other arrives knowing Ines chose to come without saying.

That is a perspective branch, and it is a modest claim next to an outcome branch. The honest thing to sell is the modest one: see the other side of the door. Not decide whether she opens it. Those are different promises. A campaign that advertises the second and delivers the first has built a button that lies.

There's a fourth element that isn't a branch at all. A product-detail control opens a six-second insert on the tote — the waxed canvas, the outer pocket, the strap — and returns to the exact frame it left, with the film continuing from there. It changes information. It changes nothing about Mara, Ines, the invitation, or the door. It's offered once, over the end card, and that placement is a decision rather than a default: the insert has to pause the film, unlike the story choice, which sits on a beat the film can spare. Anywhere earlier it either competes with the choice or interrupts the handoff.

Compare the interactive idea with the linear film

Take one moment through both: the doorbell, at twenty-six seconds in the linear.

In the forty-second cut it's a timed beat. Rain, a kitchen, a bag with two cups in it, and then a ring. The film has spent the previous sixteen seconds teaching the audience to listen for it, and the ring pays that off in about a second and a half. Nobody has to be told what it means.

An interactive version that pauses at the doorbell to ask "open the door?" spends that beat to buy a choice the viewer already made by staying. Worse, it asks a question whose answer is not in doubt: they rang, she's home. The pause converts a payoff into a menu.

So the choice doesn't go at the doorbell. It goes at eighteen seconds, on the beat where Mara closes the bag and stands a moment looking at rain. That beat is already slack — the film isn't doing anything with it — and slack is where a choice can sit without costing the film anything it was using.

The linear film, meanwhile, cannot absorb the phone. On the host path, Mara's screen is legible and unanswered. Put that in the forty-second cut and the second cup stops being a question. The audience knows someone was invited, knows nobody replied, and now reads the doorbell as a scheduling beat rather than an arrival. The suspense the second cup bought gets spent early. The phone is the object that carries knowledge, and a single-path film can't afford to carry it.

The interactive route pays for its gain. The two paths aren't the same length — roughly twenty-four seconds with Mara, twenty-eight with Ines — so the piece has two runtimes, about sixty-two and about sixty-six seconds, against the linear's forty. It needs a second location, a second actor, a soaked wardrobe state, and a wet-down. It loses the clean sixteen-second question, because the question is now split across the choice. And the shared ending has to work twice, for viewers holding different knowledge of the same twenty seconds. That last one is the real constraint, and it's a writing problem before it's a production problem.

Map enough paths to expose the real commitments

The structure, in order:

A shared opening, eighteen seconds, identical on both paths: the bread, the two sandwiches, the tote with one cup, a figure passing outside the rain-blurred window that Mara half-notices, the bag closing.

The choice, on the closed bag. Two accounts.

Host path, about twenty-four seconds: phone, no reply, second cup packed, bag carried to the door, waiting. Ends on the ring.

Ines path, about twenty-eight seconds: street, awning, the same message read and unanswered, the walk. Ends on the press of the bell.

Rejoin. Shared ending, about twenty seconds: the door, the soaked neighbour, the strawberries, the second cup handed out of the bag, the porch step, the bag between them, Ines's one line — "I didn't reply" — the end card, the detail control.

Now check what each viewer knows at the rejoin. The Mara viewer knows she packed without confirmation, and doesn't know whether the message was read. The Ines viewer knows she read it and chose to come, and doesn't know that Mara saw no reply. Both watch the same twenty seconds. When Ines says "I didn't reply," the first viewer hears a confirmation and the second hears an admission. Same line, different weight — and that's the test of whether a rejoin does work or just tidies up. The line belongs in the shared ending rather than on either path, and it is the only spoken line in the interactive build.

Rejoining isn't automatically a failure. What fails is a rejoin after a choice that changed nothing — a branch on the colour of the thermos, then everyone meets at the door with identical knowledge and a slightly longer runtime. Here the paths change what's known, so the merge has something to merge.

The commitment the map exposes is the one to decide before writing anything else: the outcome does not branch. Ines comes. She comes in every path. That's a real limit, and it's worth pricing against the alternative. Imagine a third path where Ines reads the message, thinks better of it, and walks on, and Mara eats on the porch with two cups and one person. That's a genuine outcome branch, and it costs three things. It removes the handoff, which is the product's whole role — the bag as provision for someone who hasn't confirmed. It doubles the endings, so the porch has to be written, shot, and graded twice, and only half the audience sees the strawberry payoff. And it forces the no-action default to pick a winner, which turns a declared default into a declared verdict. That third path is a possible film. It is not this build, and the spec should say so before someone puts it in a deck as included.

Write the experience when nobody acts

At eighteen seconds the choice appears. If the viewer does nothing, the piece continues on the host path and runs to the doorbell and through the shared ending without skipping anything.

Host is the default because it's continuous — the viewer has been with Mara for eighteen seconds, and staying with her doesn't require the film to introduce a stranger under its own steam — and because it delivers the handoff from the side that has been established. A hybrid default, cutting between the two accounts, isn't a third option; it's unwritten material and a second edit.

The length of the hold on the choice frame is a decision that has to be authored, written into the spec, and then checked on every surface it runs on. Don't assume a platform will hold a frame indefinitely, and don't assume it will fire a fallback. Whatever the declared hold is, the piece has to make sense at its shortest and longest values.

Two more cases the spec owes an answer to. Late entry: someone who lands at forty-five seconds, inside the shared ending, sees a cup come out of a bag and get handed to a wet person holding strawberries. That reads cold, which is lucky, but it should be checked rather than assumed — and if the campaign needs cold-entry comprehension, that's a decision with a cost, either a few words of on-screen text or a forced pre-roll of the shared opening.

Returning: the piece starts from the top. The choice is offered again, including the path not taken, and nothing is remembered — no resume position, no "you chose Mara" state, no badge. At sixty-something seconds this is cheap, and it means one viewer can see both sides in a single sitting. Resume-where-you-left-off is a state requirement with an owner, a build, and a test; if the campaign wants it, it goes in the spec rather than in the assumptions.

Separate story decisions from technical and access dependencies

The story decisions above — what changes, where the choice sits, what the default is, what the rejoin means — are all authored. Everything below belongs to someone else, and a branch diagram doesn't settle any of it.

Input method, and whether the choice works on the pointer, the remote, the keyboard, and the d-pad on the surfaces this will actually run on. The choice itself has to be legible as text and available as captioned speech, not signalled by sound alone or by colour alone.

The invitation on the phone is the piece's most fragile frame. On both the host path and the Ines path, that screen carries the fact the whole branch turns on. It has to exist as readable on-screen text at small sizes and in captions, not as a shot of a phone that only works on a big screen.

The rain does a lot of narrative work here, so both the choice and the doorbell must survive muted playback. The strawberries under the jacket have to read by shape, not by colour. Whatever gets measured about which path is taken is a data decision with a privacy owner. And the two-location, two-actor, wet-wardrobe build is a real line in a budget — the branch is not free, and the diagram won't pay for it.

Three arrangements, and they buy different things.

One film, common opening and ending, distinct middles. Shares the shoot, the ending, and one grade. It costs the linear its suspense structure: the interactive's eighteen-second opening has to withhold the second cup so the host path can spend it after the unanswered message, so a linear built on that shared opening loses the sixteen seconds of visible, unexplained cup it depends on.

An interactive companion beside an untouched forty-second cut. The linear runs exactly as approved; the interactive is a separate build for viewers who opt into it. This preserves the spot, and it means the kitchen and the porch get assembled twice.

A separately authored linear. Warranted when the interactive's payload is the gap between two people's knowledge — which, in this example, it is. A forty-second single-path film can't show Ines reading the invitation without handing the door away. So if most of the audience will only ever see the linear, that linear has to be written as its own forty seconds, not as the interactive's host path with the button removed. It also keeps the wordless porch: with no phone in its forty seconds, "I didn't reply" would refer to nothing the audience has seen, so the second cup and the strawberries carry that ending alone, as they do in the approved cut. The line stays with the interactive's shared ending, where both viewers have earned it.

What I'd do here: shoot one body of material and cut it two ways. Share the kitchen, the porch, the handoff, and the tote insert. Do not share the edit. The linear's forty seconds want the second cup visible and unexplained; the interactive's eighteen-second opening wants it withheld. Those are incompatible orders, and picking one for both is how you end up with an interactive piece whose default path is the ad you already had.

Which is the failure to watch for. If the only difference between the linear version and the interactive version is a button most viewers never press, the campaign has paid for a button. The interactive earns its difference through what participation changes — and here it changes what the viewer knows on arrival at a door that opens either way.

Every action should have its consequence stated in the spec, one line each. Stay with Mara: you'll know she packed for someone who never answered. Go with Ines: you'll know she read it and came anyway. Do nothing: you'll get Mara's account in full, nothing withheld, and you'll reach the same porch. Come back: you start at the top and may take the other side. And the interface should promise only what those lines promise — the other side of the door, not control of whether it opens.

Frequently asked questions

When does an interactive commercial need a different story from the linear film?

When the interaction redistributes what the viewer is allowed to know, not when a button is added to the same film. In the invented Wend example, the choice lets one viewer learn Mara packed without a reply and the other learn Ines read the invitation and chose to come; the ending does not branch. If the only difference between linear and interactive is a button most viewers never press, the campaign has paid for a button.

Where should the story choice sit?

On a beat the film can spare, not on the doorbell. In the example the choice goes at eighteen seconds, as Mara closes the bag and looks at the rain, because that beat is already slack. Pausing at the doorbell would spend a payoff to buy a choice the viewer already made by staying and would ask a question whose answer is not in doubt.

What happens if the viewer does nothing?

The piece continues on the host path: Mara's account in full, nothing withheld, through the doorbell and shared ending without skipping. Host is the default because it is continuous and delivers the handoff from the side already established. A hybrid default that cuts between the two accounts is not a third option; it is unwritten material and a second edit.

What does a product-detail control do?

It opens a six-second insert on the tote and returns to the exact frame, with the film continuing from there. It changes information but not Mara, Ines, the invitation, or the door. It is offered once over the end card because earlier placement would compete with the story choice or interrupt the handoff.

Shoot one body of material and cut it two ways, but do not share the edit. The linear's forty seconds want the second cup visible and unexplained; the interactive's eighteen-second opening wants it withheld so the host path can spend it after the unanswered message. Those are incompatible orders, and picking one for both can leave the interactive piece with the already-approved ad as its default path.