The feature tour is the structure that loses viewers#
Open the recorder, click through the product in menu order, and narrate each panel as you reach it: here's the dashboard, here's settings, here's the reports tab, here are the integrations. That is a feature tour, and it is the shape a demo defaults to when nobody chose one first. It fails for a reason that is structural, not a matter of polish. A tour is a list, and a list gives the viewer no reason to watch it in order. Panel two does not depend on panel one, panel four makes no promise worth waiting for, and every item is independently skippable because none of them are owed a payoff.
A story is the opposite of a list, because a story withholds. It raises a question in the first beat and settles it in the last, so the viewer stays to learn how it comes out. That forward pull is exactly what a feature tour has none of, and the retention math shows the bill. YouTube's own guidance names the failure: audiences only thin out as a video runs on, so when the top moments land late, its advice is to "consider introducing the compelling content earlier"1. A tour either buries its best moment or has none, and lets the curve decay from the first second. Wistia, across more than 13 million videos, puts sub-minute clips at about a 52% engagement rate and product videos near 50%2, so half the audience is gone by design, and a tour spends that half on housekeeping. Pick the runtime the channel wants first, since how long the video should run sets how much arc you can afford, then give it an arc.
The cost lands twice, and the second time is after the video ends. People do not store an experience as an average of its moments; they store its most intense moment and its final one. That is the peak-end rule, from Kahneman and Fredrickson's 1993 study of how memory compresses an experience down to its peak and its ending3. A feature tour has no engineered peak and no engineered ending; it runs out of panels and stops, so it leaves nothing behind to remember. Fixing that means imposing an arc: a beat that raises the stakes, a peak where the product answers them, and a close you chose on purpose. Three named frameworks do exactly that, and they differ only in how they order the beats and where they park the payoff.
Three story frameworks, one timeline#
Problem-agitate-solve, before-after-bridge, and the job-to-be-done hero are the same arc, tension then release, sequenced three different ways. Choosing among them is choosing where the viewer feels the pressure and where they get the relief. Below is one 60-second runtime read through all three at once. The seconds convert to words at about 2.5 per second, so every column is still the roughly 150-word budget the problem, walkthrough, proof script template sizes line by line; what changes across the columns is the shape, not the count.
| Beat role | Problem-agitate-solve | Before-after-bridge | Job-to-be-done hero |
|---|---|---|---|
| First frame | the painful status quo | the finished result | the viewer and their goal |
| Tension beat | agitate the recurring cost | the gap between old and new | the obstacle and its stakes |
| Body (walkthrough) | solve: the product does the job | bridge: how you cross to the result | the guide's plan, step by step |
| Close | proof, then one ask | one ask | the viewer wins, then one ask |
| Where the payoff sits | at the end | in frame one, then re-earned | the moment the viewer succeeds |
Read across any row and the three quarrel over one decision: when to show the payoff. Problem-agitate-solve makes the viewer wait for it, before-after-bridge hands it over in the first frame and spends the rest earning it back, and the hero frame delivers it as the viewer's own win. Everything else is beat budgeting, which is where the per-framework numbers below come in.
Problem, agitate, solve: make the cost visible before the fix#
Problem-agitate-solve is the classic direct-response arc, and it is what the familiar "problem, walkthrough, proof" skeleton becomes once you stop skipping the middle beat. The agitate step is the one most demos drop, and it is the one that earns the walkthrough. The problem beat states the pain; the agitate beat makes its price felt, the recurring tax of living with it, so that by the time the product appears the viewer already wants it gone.
| Beat | Budget (of 60s) | What it does |
|---|---|---|
| Problem | 0:00–0:07 | one concrete scene of the pain, product unnamed |
| Agitate | 0:07–0:15 | the cost that recurs: every week, already stale, again tomorrow |
| Solve | 0:15–0:52 | the product does the job in three visible steps |
| Proof + CTA | 0:52–0:60 | the result as a number, then exactly one action |
The eight seconds of agitation are the whole trick. On a weekly-report demo, the problem beat shows the spreadsheet; the agitate beat says the numbers are stale by Wednesday and the team argues over which copy is real. That is one honest cost, not three, and the line to hold is that agitation earns attention only while it stays true. Push it into invented dread and the viewer smells the sell. Reach for this frame when the pain is real but under-felt, which is common in outcome-framed B2B, where the buyer has normalized a cost they stopped noticing.
Before, after, bridge: open on the destination#
Before-after-bridge inverts the wait. It shows the "after," the finished result, in the opening seconds, then rewinds to the "before" long enough to make the contrast land, then spends the body as the "bridge" that connects them. It is the outcome-first cold open with a name, and it is the right arc for anywhere the payoff is visual and the surface plays muted.
| Beat | Budget (of 60s) | What it does |
|---|---|---|
| Before | 0:00–0:06 | the old way, in one shot |
| After | 0:06–0:14 | the destination on screen, done, before any how |
| Bridge | 0:14–0:52 | the three steps that connect old to new |
| CTA | 0:52–0:60 | one action, one destination |
Opening on the after is why this frame survives autoplay. A viewer who leaves at four seconds still saw where the product ends up, and a viewer who stays now watches the bridge to confirm a result they have already been shown, which is a stronger reason to keep watching than curiosity about a feature. That is also why this is the safest structure for the first frames a muted hero shows; it makes the opening five seconds carry the whole pitch as silent motion. Use before-after-bridge when the contrast reads at a glance: a cluttered board becoming one clean view, a red build turning green.
The job-to-be-done hero: cast the viewer, not the product#
The third frame changes who the story is about. Clayton Christensen's jobs-to-be-done lens holds that customers do not buy a product so much as "hire" it to make progress in a particular circumstance4. Donald Miller's StoryBrand pushes the same idea into structure: the customer is the hero, and the brand is the guide who hands them a plan, not the hero itself5. A hero-frame demo casts the viewer as the protagonist with a job to finish, and the product enters as the tool that makes the finishing possible.
| Beat | Budget (of 60s) | What it does |
|---|---|---|
| Character + job | 0:00–0:06 | the viewer and what they are trying to get done |
| Stakes | 0:06–0:13 | what is in the way and what it costs them |
| Guide + plan | 0:13–0:48 | the product enters as helper and hands over a plan |
| Success | 0:48–0:55 | the job done, framed as the viewer's win |
| CTA | 0:55–0:60 | one next step |
The register gives it away: the narration is second person, "you connect your data, you ask the question, you get the chart," because the viewer is meant to picture their own hands on the interface. That makes it the natural frame for onboarding and product-led surfaces, where the sale is the viewer imagining themselves succeeding in the next minute. It also inverts cleanly for a sales-led cut, where the hero becomes the buyer's organization and the plan becomes an outcome a champion can carry into a room; which register a B2B versus a PLG audience wants is set by who has to say yes. The failure to avoid is the common one: letting the product be the hero, all capability and no protagonist, which is just a feature tour wearing a story's clothes.
Pick the frame by what the payoff looks like#
The choice is not taste, it is a reading of your payoff and your surface. One honest cost the buyer has stopped noticing wants problem-agitate-solve. A result that reads in a single frame wants before-after-bridge. A viewer who has to see themselves doing the job wants the hero frame. The table collapses it.
| Framework | Reach for it when | Payoff placement | The failure to avoid |
|---|---|---|---|
| Problem-agitate-solve | the pain is expensive but under-felt; outcome-framed B2B | at the end | agitation curdling into fear; keep it to one true cost |
| Before-after-bridge | a muted autoplay hero or social clip; the result is visually obvious | frame one | a "before" that overstays; cap it at one beat |
| Job-to-be-done hero | onboarding, PLG, self-serve; the viewer must picture themselves | the viewer's win | letting the product be the hero instead of the guide |
Here is the part the framework debate usually misses: these are not three recordings. They are one take of one workflow, re-sequenced and re-voiced, the same clicks with a different beat order and a different narration laid over them. Testing whether the before-after cut beats the problem-first cut should cost an edit, not a re-shoot, and it only does when the demo is a spec instead of a performance. Our own engine, aidemo, is built on exactly that premise, and it comes with real limits: it drives a real browser and nothing else, a coding agent writes the storyboard as text rather than a person cutting clips, and it offers no visual timeline for tweaking a beat by hand. Decide the frame before you record, because the beat order is the earliest and cheapest choice in the whole production playbook, and reusable storyboard patterns are how a chosen frame becomes a thing you can re-cut instead of re-perform.
Sources#
- YouTube Help — Audience retention (intro percentage and drop-off)
- Wistia — How to Choose the Right Marketing Video Length (2026 State of Video)
- Kahneman & Fredrickson — the peak-end rule (1993)
- Christensen, Hall, Dillon & Duncan — Know Your Customers' Jobs to Be Done (HBR, 2016)
- Donald Miller / StoryBrand — The Customer Is the Hero
- Tools for Clear Speech, Baruch College — Speaking Rate
FAQ#
What is the best storytelling framework for a product demo?#
There is no single best one; there is a best fit for your payoff. Reach for problem-agitate-solve when the pain is real but the buyer has stopped noticing its cost, before-after-bridge when the result reads in a single frame and the surface plays muted, and the job-to-be-done hero frame when the viewer has to picture themselves finishing the task, as in onboarding. All three impose the same tension-then-release arc a feature tour lacks; they only disagree about where to place the payoff.
What is the difference between problem-agitate-solve and before-after-bridge?#
Where the payoff sits. Problem-agitate-solve withholds the result: it opens on the pain, agitates the recurring cost, and delivers the solution at the end, so the viewer waits for relief. Before-after-bridge opens on the "after," the finished result in the first frame, then rewinds to the "before" and spends the body bridging the two. Because it front-loads the payoff, before-after-bridge is the safer choice for a muted autoplay hero, where a viewer who leaves at four seconds still saw the destination.
Why does a feature tour lose viewers?#
A feature tour is a list, and a list has no arc, so nothing pulls a viewer from one panel to the next. Retention decays from the first second because no beat is owed a payoff, which matches YouTube's finding that audiences only shrink over a video's length. It also fails the peak-end rule: with no engineered peak and no chosen ending, the video runs out of features and stops, leaving the viewer nothing to remember. A named story frame fixes both by raising stakes, resolving them at a peak, and closing on purpose.
