Build a Treatment With Stills, Clips, and Text Without Three Competing Reading Speeds
Build a Treatment With Stills, Clips, and Text Without Three Competing Reading Speeds
A treatment passage that puts a looping clip, a dense paragraph, and a comparison still on the same screen has handed its recipient three jobs and no order in which to do them. The recipient does what anyone does with three simultaneous demands: reads half a sentence, glances at the clip, scrolls, and comes away with an impression that is mostly about being busy.
The remedy is not fewer media. It is an order. One thing to read, one thing to watch by choice, one thing to inspect afterward — plus a stated way through when the clip never plays.
That order belongs to the document, not the ad. In the finished spot the reveal may land late and fast, and the audience gets no say in the sequence. In the treatment, the recipient decides when the clip starts. Building the document as a miniature of the commercial fights the one advantage the document has.
Give each medium a different piece of the explanation
Take one passage — not the whole deck — and write down what it has to make the recipient understand. Then assign each item by asking what kind of thing it is.
- A still holds a relationship still long enough to inspect: two states side by side, a proportion, a silhouette, a before and after that can be compared at leisure.
- A clip carries change through time: a motion, a cut, an order of events, a rhythm. It cannot be inspected. It has to be played.
- Text carries the reason. Why this reference, what it replaces, which constraint the client set, what the other option would have cost.
Those three jobs are genuinely different, and that difference is the opportunity. They turn into competing reading speeds only when two of them are asked to do the same job.
Here is the test. If a claim appears in the caption, in the still's lighting, and in the clip's score, it is one claim wearing three costumes, and two of the costumes are atmosphere. Cut them.
The example
The rest of this article works through one fictional passage, so the decisions stay concrete. The product, the client, and the clip are invented for the demonstration.
The passage covers a single beat of a proposed thirty-second spot for a hard-shell carry-on: the reveal. Three claims have to land.
- The reveal comes late. The traveler moves through a terminal for most of the spot before the case is seen clearly — it lands at about twenty-two seconds of thirty.
- The reveal is one motion. The front panel travels up and locks. There is no cut.
- The reason for the reference. The motion should read as a mechanism rather than a trick, because the client's claim is that the panel carries the load instead of decorating it.
Assignment: claim two goes to the clip, because sliding and cutting are distinguishable only in time. Claim three goes to text, because it is an argument about intent. Claim one is text as well — "lands at 0:22 of 0:30" is a number, and a number is faster to read than to demonstrate.
The still gets the inspection job: closed state and open state, same angle, same distance, so the recipient can see what the panel's travel does to the case's outline. That is a relationship, and relationships reward a long look.
The order that follows: one line of text stating where the beat falls, the poster frame, the viewing question, a play control, and the endpoint comparison for after playback.
What gets removed: a terminal photograph that sets mood, its repetition in the copy as "unhurried airport life," and the first two seconds of the clip showing the same thing. Keep one.
The reading order is not the spot's order
In the spot, the reveal arrives late and passes quickly. In the treatment, the recipient should meet the poster frame and the viewing question before the play control, and the still — the endpoint comparison — after it. The document's order serves comprehension. The spot's order serves effect. They will not match, and they should not be forced to.
Tell the reader what to notice before playback
Put the viewing instruction above the clip and above the play control. Under a moving image, the recipient meets the motion and the explanation at the same moment — exactly the collision you were trying to avoid.
Make the instruction name what changes over time, because that is the part only the clip can supply.
Watch the panel: it travels up and locks. There is no cut.
That sentence does one job. It says where to look and what to check. It does not say what to feel.
A beautiful moment of transformation that captures the brand's effortless confidence.
That is not an instruction. It is a verdict delivered before the evidence, and a recipient who is being managed can tell. Treatments get approved or questioned by people who are used to being sold to; a pre-loaded reaction reads as a reason to be suspicious of everything nearby.
The failure in the other direction is just as common: a paragraph describing every moment of a six-second clip. It looks thorough. It converts watching into auditing, and the recipient ends up checking the description against the footage instead of seeing the footage.
One question, or one instruction. If you need two, they probably belong to two different clips.
Keep the next task from starting too soon
A recipient should be able to finish reading before anything moves, and finish watching before being asked to compare.
Four practical means:
- Turn autoplay off if the export allows it. A poster frame and a play control let the reader arrive at the clip rather than get caught by it mid-sentence.
- Pick a poster frame that is a question, not the answer. For the reveal, that is the panel partway up — not the finished silhouette, which gives away the result before the viewing instruction has been read.
- Place a return point after playback. In the example, the endpoint comparison carries a short label that marks it as the thing you look at once you have seen the motion, not a second image competing with it.
- Treat replay as support. A recipient may want a second look at the lock. That is a good reason to replay. Needing three passes to understand the first one means the viewing question was doing too little work.
Design a useful skipped-media route
Ask what remains understandable when the recipient never presses play. Some will not: someone skimming before a meeting, someone on a phone in a taxi, someone who reads faster than they watch. And some deliveries simply will not play the clip, for reasons that have nothing to do with anyone's preference.
Three things should survive the skip: the claim, the reference's intended role, and a clearly labeled description or a set of selected states.
For deciding what to write in that route, the W3C Web Accessibility Initiative's planning guidance for audio and video is a reasonable starting point, because it sorts alternatives by what the media actually carries — spoken content, sound, or visual information (page checked September 8, 2026). The useful part is that distinction: what you write down changes depending on whether the essential content is something said, something heard, or something seen. Two limits are worth stating plainly. That guidance supports the distinction between kinds of alternatives. It does not endorse the reading sequence described here, and it is not a compliance verdict for a treatment.
In the example, the clip's essential content is visual — a soft mechanical click marks the lock, but the stop reads without it — and the route reads:
If you skip the clip: the panel travels upward as one continuous motion and locks. It does not cut. The two states below are the start and end of that travel.
That preserves the claim and the mechanism. It does not preserve duration or pace, and no still can. So do not call it an equivalent. Call it a fallback that keeps the decision possible. If the client's decision turns on whether the motion feels fast or deliberate, they need the timing, and the honest move is to say so rather than let two stills stand in for it.
Read the passage in its actual delivery conditions
Walk the passage through from the recipient's side rather than the author's. Four conditions are worth checking every time.
A small display. The two-up comparison in the example is side by side at desktop width and stacked at phone width. Stacking changes the order in which the two states are read, so check that the relationship still holds when the images sit one above the other, and that the caption stays with the image it describes.
Unavailable audio. In the invented example, the lock is marked by a soft mechanical click. With the sound gone, the panel's stop still reads visually — worth knowing before anyone leans on the click. If a beat is carried only by sound, the no-play route has to carry it in words, and it should say that words are what is carrying it.
Failed playback. A poster and a control that does nothing. Check whether the viewing question still has an answer nearby, and whether the surrounding text still makes sense with no image at all.
Interruption. Someone reads half the passage, leaves for an hour, and comes back. Does anything restate what to read, play, and compare?
When a walkthrough turns up a sequencing problem, fix it by reassigning explanatory work — not by adding captions, instructions, or a second copy of the same image. If the endpoint comparison is unclear, the relationship may have been temporal all along and belongs in the clip. If the clip is unclear, it may be carrying an argument that text should have carried from the start.
The checks above are a checklist, not results. The layouts, the clip, and the walkthrough for this example have not been produced, and nobody has measured whether a sequence like this reduces anyone's cognitive load. The narrower claim is the one worth making: this passage specifies what to read, play, and compare, and it says what is lost when the clip is skipped.
What the recipient should be able to say
After one pass through the passage, the recipient should be able to produce four sentences without going back.
I read the reason the reveal is a mechanism. I chose to play the six-second clip. I compared the closed and open states. I skipped nothing — or I skipped the clip and lost the timing.
If the third sentence is "I compared nothing," the comparison still was not doing inspection work. If the fourth is "I lost nothing," the clip was decoration.
Test the passage that way and most of the editing decides itself. Then run the same four questions on the next passage. Its answers will be different, because it has different claims to make.
Frequently asked questions
How should I decide whether a claim belongs in text, a still, or a clip?
Assign by the kind of explanation it carries. A still holds a relationship still long enough to inspect: two states side by side, a proportion, a silhouette, a before and after. A clip carries change through time: motion, a cut, an order of events, rhythm. Text carries the reason: why this reference, what it replaces, which constraint the client set, what the other option would have cost. If the same claim appears in the caption, the still's lighting, and the clip's score, cut two of the three.
Should the treatment's reading order match the finished commercial's order?
No. The document's order serves comprehension; the spot's order serves effect. In the invented carry-on example, the spot's reveal arrives late and passes quickly, but the treatment should let the recipient meet the poster frame and viewing question before the play control, then the endpoint-comparison still after playback. They should not be forced to match.
What should a viewer be told before playing a clip?
Put the viewing instruction above the clip and above the play control. Make it name what changes over time, because only the clip can supply that. The example is: 'Watch the panel: it travels up and locks. There is no cut.' Avoid a verdict delivered before the evidence, such as calling it a beautiful transformation, and avoid a moment-by-moment description that turns watching into auditing. One question or one instruction.
What should survive if someone skips the clip?
Three things: the claim, the reference's intended role, and a clearly labeled description or a set of selected states. The body cites W3C Web Accessibility Initiative planning guidance because it sorts alternatives by what the media carries—spoken content, sound, or visual information—but that guidance does not endorse the reading sequence described and is not a compliance verdict for a treatment. In the example, the fallback preserves the claim and mechanism, not duration or pace; no still can. Do not call it an equivalent. If the decision turns on whether the motion feels fast or deliberate, say the timing is needed.
How can I test whether the passage works?
After one pass, the recipient should be able to say: I read the reason the reveal is a mechanism; I chose to play the six-second clip; I compared the closed and open states; I skipped nothing—or I skipped the clip and lost the timing. If the third sentence is 'I compared nothing,' the comparison still was not doing inspection work. If the fourth is 'I lost nothing,' the clip was decoration. Also walk the passage through small display, unavailable audio, failed playback, and interruption; fix problems by reassigning explanatory work, not by adding captions, instructions, or a second copy of the same image.