A Feature Is Not a Selling Angle: One SaaS, Twelve Different Videos

You shipped the thing. Now you need a video for the launch, and you do the reasonable engineer move: open a doc and list what the product does. Groups duplicate errors. Sends a Slack alert. Supports source maps. Installs in five minutes.

Then you record it, watch it back, and something is wrong. Every sentence is true. Nothing in it makes you want the product. You have made a changelog with music.

The problem is not your delivery or your screen recording. It is that a feature list and a selling angle are different kinds of statement, and only one of them can carry a video.

The test: can someone disagree with it?

A feature is a fact about your software. An angle is a claim about your viewer's situation.

Facts are unarguable, which sounds like a strength and is actually the problem. Nobody watches "groups duplicate stack traces" and thinks no it doesn't. Nobody thinks anything at all. There is no position to take, so the viewer takes none, and a viewer with no position scrolls.

Now try this opening instead:

You found out about the outage from a customer. Again.

Someone can disagree with that. A team with real on-call rotation watches it and thinks not us. And the person it describes goes quiet for a second, because it happened last month and they still think about it.

That is the whole test. If a reasonable person cannot disagree with your opening line, it is not an angle. Disagreement is possible only when the sentence is about the viewer rather than the product — and a sentence about the viewer is the only kind that earns the next three seconds.

Your features are still doing work in this version. "Again" is only credible because the product actually detects the outage. But the feature has moved out of the argument and into the evidence, which is where it belongs.

One product, twelve angles

Take a small error-tracking SaaS for teams too small to have an SRE. Same five features throughout. Twelve genuinely different videos.

Angle	Opening line	What it bets on
Pain	"You found out about the outage from a customer. Again."	The viewer has lived this specific humiliation
Cost of inaction	"Every month you put this off costs about one engineer-week of debugging blind."	Nothing is the real competitor, so make nothing expensive
Before / after	"Monday: grep through four log files. Tuesday: open one screen."	The gap argues by itself if both halves are honest
Comparison	"Your current setup is console.log and hope. Here's the one thing that changes."	Names the real alternative — usually not a competitor
Mistake	"Most small teams log everything and alert on nothing."	You have earned the right to correct the viewer
Myth-busting	"Error tracking is not a thing you add once you're big."	One belief is blocking signup and can be replaced
Use case	"A two-person team shipping on Fridays. Here's their Friday now."	Recognition beats reach for a horizontal tool
How it works	"One import line. Here is what happens on the next crash."	Interest already exists and needs to become belief
Feature spotlight	"Four hundred errors. Three actual bugs. Here's how it knows."	One capability is strong enough to hold the runtime alone
Social proof	"Six hundred teams under ten people run this in production."	Specific evidence exists — vague proof is worse than none
Testimonial	"I stopped checking the dashboard on weekends." — a real user	A customer's claim outweighs yours, if it stays rough
Statistic	"The median bug reaches a user 40 minutes before it reaches the team."	One sourceable number is more memorable than a paragraph

Read them straight down. Nothing about the software changed. Twelve different arguments came out of it, and they are not interchangeable — the statistic angle and the testimonial angle are aimed at different people in different moods, and half of them would be wrong for any given launch. That is the point. An angle that fits every situation is not an angle; it is a tagline.

Each of those links goes to a full breakdown of when the angle works, when it backfires, and what its script structure looks like. Or start at the angle library if you want to browse.

An angle is still not a video

A line is a promise that a video has to keep. Take the cost-of-inaction one and put it on a timeline:

Shot 1 — the number (0-4s). On screen: your own repo, a PR merged three weeks ago. Voice or caption: "This bug shipped on the 3rd. A user hit it on the 4th. You saw it on the 22nd."

Shot 2 — the arithmetic (4-9s). On screen: a simple counter, 18 days. Caption: "Eighteen days of one broken signup path. At your conversion rate, that is a number you can actually calculate."

Shot 3 — the turn (9-13s). On screen: the real product, one alert firing in Slack. Caption: "Or you find out at 14:32 on the 3rd." This is the only beat that is about you, and it needs no more than one line.

Shot 4 — the ask (13-18s). On screen: the install line, five seconds of it working. Caption: "One import. Free until you have something to track."

Four shots, under twenty seconds, and every one of them is a screen you already have. Most of this footage is a walkthrough you recorded for something else — there is a reusable path from a raw screen recording to a launch video that assumes exactly that starting point.

Notice what happened to the feature list. "Slack alerts" and "five-minute install" are both in there, in shots 3 and 4. They never had to be cut. They just could not go first.

You need three, not twelve

Twelve is the menu, not the order. For a launch, three is usually right — and the three have to disagree with each other.

Here is the failure mode, and it is common enough to be worth a diagnostic: you make three videos, they perform identically badly, and you conclude video does not work for your product. Open the three scripts and read only the first lines. If you could swap them between videos and no one would notice, you did not make three angles. You made one angle in three fonts.

Real separation looks like picking one angle for people who do not know the category exists (pain or statistic), one for people already shopping and comparing (comparison or how it works), and one for people who believe you but have not moved (cost of inaction or testimonial). Three audiences, three moods, three arguments. Now the results tell you something, because the variable you changed was the argument rather than the background music.

Working out those three from a product page is a mechanical enough pass that we built a generator for it — but the reasoning is the same whether a tool does the first draft or you do.

The short version

Your feature list is not wasted. It is your evidence, and evidence is load-bearing — it is what stops the angle from being a nice sentence with nothing behind it.

It is just not the argument. The argument is a claim about your viewer that some of them will reject, and writing one means giving up the safety of saying only true things about your software.

Test every opening line you write against one question: could a reasonable person watch this and think that's not me? If nobody can, nobody will think that's exactly me either.