Write a Clickable Commercial Around Action, Response, and Return
Write a Clickable Commercial Around Action, Response, and Return
Every interactive treatment contains a sentence like this one:
Hotspots let viewers explore the details of the workbench.
Six words of interface, no view of an actual experience. Nobody reading it can say what a viewer does, what they see afterward, or how they get out again. The client can't approve it. The editor can't build it. And if it ever ships, someone will press the wrong thing twice, get nothing back, and stop.
The line is fixable, and the fix isn't more atmosphere. Describe the cycle: what signals that something can be pressed, what the viewer does, what visibly changes, and how they get back. Then write the two paths that get skipped — the wrong press and the doing-nothing press. That order matters. A clickable commercial needs an understandable action-and-feedback cycle before it needs a branching story, because a branch is a decision about narrative and a cycle is a description of an experience. You can't pitch the second convincingly without the first.
The example I'll use throughout is invented: a 40-second film of someone repairing something at a crowded workbench. It is not built, drawn, or tested, and I'm not claiming it would work on any platform. It's here because a described cycle is much easier to argue about than a principle, and because a client can look at it and say "cut the clamp."
Write the invitation and the action
An action needs an invitation. Something on screen has to look pressable — a label, a marker, a ring that appears at the right moment — and the treatment should say what it is, where it sits, and when it shows up. "Hotspots" is a category, not a description. A small marked control beside the lamp base, visible from 00:07 to 00:21 is a description. So is a standing button in the lower corner that never leaves the frame. The art director needs the difference, because the invitation is where interactive work usually fails: a faint circle nobody notices, or a badge so loud it reads as a display ad.
Then name the operation, precisely. The W3C Web Accessibility Initiative's button pattern draws a line between a control that acts on the current page and a link that navigates elsewhere (pattern documentation, inspected September 2026). Advertising language blurs exactly that line, because "click-through" already means something else to us: the click that leaves the ad and opens a landing page. That's navigation. It ends the film, and the viewer is somewhere new with a different job.
A control that reveals the underside of the lamp clamp keeps the viewer where they are. Two different operations, frequently rendered as identical-looking buttons. If the treatment says "interactive" and doesn't say which one it means, someone downstream will guess, and the guess becomes the build.
Write the response as something you could point to
Precise questions save a lot of trouble: after activation, what is on screen that wasn't there before? A layer uncovered, a camera push-in, a cutaway, a paragraph of copy sliding in, the next scene unlocked. And how does the viewer know the press registered — that the thing they did produced the thing they're seeing?
"The camera holds, then cuts to a four-second view of the clamp's underside, where a stamped batch number is legible." That's checkable. "An immersive exploration of product detail" is not. Neither is "the interface responds to the user."
Feedback and response are not the same job. The response is the content change. The feedback is the small acknowledgment that answers did that work? — a marker that pulses, a label that changes, a cursor state, a beat of silence before the cut. On a slow connection or a hesitant press, feedback carries the whole experience.
This is also the point to be disciplined about promises. Personalization, saved state, a viewer's earlier choice remembered on a return visit — those are product capabilities, not treatment adjectives. If nobody has developed one, don't write it. A treatment is a promise, and the client will hold you to the sentence, not the intention behind it.
Write the wrong press, the way back, and the do-nothing path
The wrong press is the most commonly omitted path in interactive advertising. Any input surface has more than one possible input, and the viewer will find them. In the workbench film, suppose the marked control is the clamp and somebody presses the jar of screws instead. If nothing happens, the viewer cannot tell whether they missed the target or the thing is broken. Both readings lose them.
The fictional treatment answers neutrally: the clamp marker pulses twice and one line of text appears — Only the clamp opens. Short, calm, no reward and no punishment. It teaches the rule in one try, and the film keeps running, so there's no dead end to climb out of. The same answer comes back every time, which matters. An occasional different response teaches the viewer to hunt for a secret that isn't there.
That design costs something, and the treatment should own it. Making the whole bench an input surface means production listens on the entire frame rather than one hotspot. Cheaper versions exist, and one of them is simply making only the clamp pressable — but then the rest of the image is inert, and inert pixels feel broken.
Then the way back. If the action changed the view, what un-changes it? A second press of the same control, a separate Back to the bench, a tap anywhere. Say which, and say what state the viewer lands in. If there's no return — a one-way reveal — say that too, and then account for it, because the viewer has left the film's default state and stays gone.
One more piece of the return, easy to forget: the viewer who pauses and then walks away. An interaction that holds the film indefinitely is a place people can get stranded. The workbench treatment resumes on its own after eight seconds of stillness. That takes a choice out of the viewer's hands, which is a real cost, and the treatment should say so — and name the state it leaves behind — rather than let either be discovered in the edit.
Finally, the path where nothing happens at all. Write the film as if nobody touches anything: it runs its length, reaches its ending, and the ad's argument still lands. If the reveal is required to understand the payoff, then the do-nothing route is broken, and the interactive version is really a different film — one where the ending is behind a press. In the workbench film the reveal is a supplement: the clamp's underside is interesting, and the story doesn't need it. That's what keeps the no-action route whole.
Write the input you're assuming
A treatment that says "hover to reveal" has quietly chosen a device. Hover has no touch equivalent. A swipe has no keyboard equivalent. The proposal should name its assumed input and at least one alternative, in the same breath, before anyone prices the build.
The W3C button pattern is the right list of questions to answer, and it is not a certificate. It covers accessible naming, keyboard activation by Enter and Space, and where focus goes after an action (documentation text, inspected September 2026). Turn each into a line your treatment answers: what is the control called? Does Enter work as well as a click? After activation, does focus stay on the control, move into the new content, or go nowhere? If focus stays put and the control itself toggles, the same key reverses the action, which is a satisfying small loop — and you only get it by deciding.
Naming deserves its own line, because it's easy to fudge. A visible label can be short, since it sits next to a lamp in a crowded frame. The accessible name should say what the control does. Where those diverge, the treatment should flag how a viewer using voice control would ask for it. And if the reveal carries a claim the ad actually depends on — the batch number is the message — then that claim needs a route that doesn't require anyone to press anything. Otherwise the ad's point is conditional on an interaction.
None of this makes an unbuilt commercial accessible. It produces a requirement list that someone can test against during the build. Testing is the only thing that establishes that the cycle works, and it hasn't happened here.
Decide whether the story changes
A detail reveal and a narrative branch look similar in a treatment and behave nothing alike.
The reveal is a detour. The film pauses, shows something, and returns to the frame it left, so the viewer loses no time and misses nothing. Its return is genuinely a return. A branch is a substitution: in a second fictional version of the same film, the marker offers Hand her the wrench and Hand her the tape, and whichever the viewer picks is what she uses in the next shot. There is no Back to the bench for that. Pressing back would mean un-happening something, and the film would have to replay material the viewer has already watched.
That difference is what forces a different account. The reveal needs a cutaway and one return state. The branch needs both continuations written, both shot and edited, an account of how they rejoin the film — or an admission that the film ends two different ways — and a decision about what the viewer who does nothing sees, since that default is the version most of the audience gets. It also needs a second argument. If the two versions say the same thing, a whole scene has been bought for nothing. If they say different things, the ad now has two positions, and somebody has to be comfortable with both.
And they need to look different. If the clamp control toggles but the wrench control commits, and the two are drawn identically, the viewer learns the rule by getting caught. If a proposal mixes both kinds of control in one film, that risk belongs in the treatment.
The test before all of this is short. If you can't write what changes and how the viewer gets back, remove the control. An interaction with no useful consequence still costs a build, a quality pass, a line of copy, and a beat of attention that a 40-second film doesn't have to spare.
The same sequence, written for a treatment
The paragraph below is a paper description of a fictional sequence, in the form a treatment would carry it. Nothing here has been built, and the accessibility points are requirements to check in production, not a claim about a delivered experience.
Bench — default. The film runs. At 00:07 a small marked control appears beside the lamp base, labeled Under the clamp. It stays until 00:21. Proposed inputs: pointer, tap, Enter, Space. The control's accessible name: "Show the underside of the clamp." (Open item: confirm how a viewer using voice control names this, since the visible label is shorter.)
Action. Pressing the control holds the film on its current frame and cuts to a four-second view of the clamp's underside, where a stamped batch number is legible. Focus remains on the control, whose label becomes Back to the bench. Pressing it again restores the held frame and the film resumes from that frame. If no input arrives for eight seconds, the film resumes on its own: the cutaway closes, the held frame gives way to the running film, and the label reverts to Under the clamp, with the control still on screen until its 00:21 window ends. A press after that auto-resume reopens the underside exactly as before, and the eight-second timer starts again. The reveal adds time to the sitting, not to the film: the film itself stays 40 seconds.
Wrong press. Any press outside the control during the window produces one response — the marker pulses twice and a single line appears: Only the clamp opens. The film keeps running. The answer is the same every time.
No action. If the viewer never presses anything, the marker fades at 00:21 and the film plays through to the ending it always had. Nothing in the reveal is needed to follow the story.
Frequently asked questions
What is wrong with a treatment line like “Hotspots let viewers explore the details”?
It names a category of interface rather than a describable experience. A reader cannot say what signals that something can be pressed, what the viewer does, what visibly changes, or how the viewer gets back. The fix is not more atmosphere but a described action-and-feedback cycle.
What is the difference between response and feedback in an interactive commercial?
Response is the content change: what is on screen after activation, such as a layer uncovered, a camera push-in, a cutaway, or copy sliding in. Feedback is the small acknowledgment that answers whether the press registered, such as a marker that pulses, a label that changes, a cursor state, or a beat of silence before the cut. On a slow connection or a hesitant press, feedback carries the whole experience.
Why must a treatment write the wrong-press path and the do-nothing path?
The wrong press is the most commonly omitted path. If pressing something other than the control does nothing, the viewer cannot tell whether they missed the target or the thing is broken. A neutral, repeated response teaches the rule in one try without a dead end. The do-nothing path must be written as if nobody touches anything: the film runs its length, reaches its ending, and the ad’s argument still lands. If the reveal is required to understand the payoff, the interactive version is really a different film.
What should be decided about input and accessibility before build?
The treatment should name its assumed input and at least one alternative. Hover has no touch equivalent, and a swipe has no keyboard equivalent. The W3C button pattern raises questions about accessible naming, keyboard activation by Enter and Space, and where focus goes after an action. It is a list of requirements to test during the build, not a certificate, and it does not by itself make an unbuilt commercial accessible.
How does a detail reveal differ from a narrative branch?
A reveal is a detour: the film pauses, shows something, and returns to the frame it left. A branch is a substitution: the viewer’s choice changes what happens next, so there is no simple return without un-happening something. A branch needs both continuations written, shot, and edited; an account of how they rejoin or an admission that the film ends two ways; and a decision about what the viewer who does nothing sees. If you cannot write what changes and how the viewer gets back, the control should be removed.