Social-Media Teaser
You post a short ad for a thing that doesn't exist yet, aimed at the people you think want it, and you watch who reaches for it. No product, no landing page worth the name. Just a post and a promise.
It's two experiments riding in one. The post tells you whether anyone wants the idea, and which way of describing it pulls hardest. Demand and messaging, read off the same small budget. (Demand and desirability are the same thing here, the question of whether people want it.)
The question it answers
This one reads desirability and messaging at the same time. Desirability is the do-people-want-it question. Does the market care, and which words make it care?
Most teams argue for a week about how to pitch the thing, then build it pitched the way the loudest person in the room liked. The teaser settles that argument with strangers instead of opinions. You don't pick the headline. You run three and let the spend pick.
It does not test usability or feasibility, the other two questions on the list. Feasibility is can-we-build-it. Usability is can-people-actually-use-it once it's built. A teaser can tell you which angle pulled. It can't tell you any of that. Those are different rungs, and you climb to them later if the teaser comes back yes.
Think of the experiments as a ladder. At the bottom sit the cheapest tests, the ones that give the weakest signal for the least money. Higher up they cost more and prove more. The taxonomy underneath this, a catalog that sorts experiments by what they test and how much they cost, comes from Testing Business Ideas by David Bland and Alex Osterwalder, who lay out 44 of them end to end. The teaser is one of the cheapest in the book, which is why it sits near the bottom of the ladder.
How to run it cheaply
The minimum version is three posts and a small spend. Here's the whole recipe.
- Write the success number down first, before you launch. This is the one rule the whole catalog leans on. The number that closes the demand question is the click-through rate, the share of people who saw the ad and clicked it. A rough rule of thumb. Aim for around 1 to 2%, and if you can collect it, a sign-up conversion above 5% on whoever clicks confirms the read (of the people who clicked, the share who then signed up). Those are starting numbers, not research findings. Set yours, for your audience, before you spend a dollar. A target you set after seeing the results isn't a target.
- Write three versions of one ad. Same product, three stories. Each leads with a different reason someone would want it. You're not testing three products. You're testing three ways of describing one.
- Aim each at the audience you believe wants it, and put a small spend behind each. Equal spend, same window, so the only thing that varies is the words.
- Read which angle pulled, and read whether any of them cleared the bar you set. The winner isn't the one you liked. It's the one strangers clicked.
That's it. No supply chain, no landing page, no code. A few days and a small budget.
A worked example
Picture a meal-kit service aimed at people who work twelve-hour hospital shifts. Pre-portioned, heat-in-ten, dietitian-approved. This is invented, like every company and number below, to make the mechanics concrete. The method is real. The result isn't a claim.
You're earlier than a landing page here, so you write three versions of one ad.
One leads with time: "dinner in ten minutes after a twelve-hour shift." One leads with health: "real food on the worst-eating schedule there is." One leads with guilt relief: "stop apologizing to yourself for the vending machine." Same product. Three stories.
Before launch you write the bar down, then run a small spend behind each, equal money, same week.
The guilt-relief ad pulls a 2.3% click-through. The other two limp in under 1%. And the people who clicked the guilt-relief ad signed up at a rate that cleared the bar you wrote down, so the secondary number agrees with the first. (All illustrative.)
Now look at what one small budget bought you. Two answers, not one. People want it, and you can name which words make them want it. The winning angle, guilt relief, goes on the landing page you build next, instead of on whichever pitch the room voted for.
You didn't build a thing to find out if the thing was worth building. You sold the idea with a sentence, three different ways, and let strangers tell you which sentence was true.
When to reach for it, and when not
Reach for it when you want a fast, cheap read on demand and messaging together, before you've built a landing page worth the name. It's a bottom-of-the-ladder move, no engineering. When you're still arguing about how to describe the thing, this is the experiment that ends the argument.
Don't reach for it when the message is already settled and the only open question is who to aim at. That's the online ad test next door, where you hold the words steady and vary the audience instead. Most of the time the teaser and the ad test are the same machine pointed at a different thing you change. If your teaser is already changing one thing and reading one number, you don't need both.
And don't reach for it when you need more than a click. A click is the cheapest signal there is. People click to make a notification go away. If you need to know whether the interest survives a real cost, climb a rung. The landing-page test asks for an email or a reservation. Higher up, the pre-sell asks for a card. The teaser is the first knock on the door, not proof anyone will walk through it.
Pick the cheapest experiment that answers the exact question you have. For "do they want it, and what should I say," the teaser is usually the cheapest thing on the shelf. Run it first. Climb only when the question forces you to.