How do you know if your digital product will sell?

This question usually gets answered with numbers, and the numbers are usually made up. Search volume from a tool that has never seen your product. A revenue figure from a stranger on YouTube. A category that is "booming". None of it tells you whether one specific file, at one specific price, described in your specific words, will get a stranger to enter their card details.

So this post does not try to predict. It tries to make the test small.

Can you know before you build it?

No. You can only lower the cost of being wrong.

That is worth sitting with for a second, because most of the advice in this niche is dressed up as prediction. Validation frameworks, demand scores, niche-picking checklists. They all move in the same direction: they give you a reason to feel confident, and confidence is not the same thing as information.

The reason prediction fails here is that a digital product is not a commodity with a market price. Two people can sell the same kind of template, at the same price, to the same audience, and get completely different results, because the offer, the wording, the preview images and the trust behind the account are all part of the product. None of those are in anybody's dataset.

This is my reasoning rather than a finding, and I would rather say so. What follows from it is practical though: if you cannot predict, the only sensible move is to shrink the thing you are risking. Less build time, smaller scope, earlier price.

What counts as evidence that someone wants this?

Money first, time second, words last. That ordering is the whole idea.

A stranger paying is the strongest signal you can get, and it tests your price at the same time as your idea. A stranger is someone with no social reason to be kind to you.

Somebody spending time is the next best. Handing over an email address for a sample, sitting through a six minute walkthrough, replying with a detailed question about whether it works in their setup. It costs them something, just not much.

Somebody saying they would buy it is worth close to nothing, and it is the signal most people build on, because it is the easiest one to collect. It costs the person nothing to say yes, and it costs them something awkward to say no to your face.

One more filter on all three: evidence from people who already know you tells you about your relationship, not about your product. Your group chat is not a market. If the only interest you can find is from people who would feel rude ignoring you, you have not tested anything yet.

What only feels like evidence?

Five things that look like proof and are not.

• Engagement on a post about the topic. A saved reel about wedding planning is interest in wedding planning. It is not interest in your $29 file.
• A keyword tool number. Search volume measures how many people type a phrase. It does not measure whether any of them were going to buy anything, or whether they would buy yours instead of the free version in the first result.
• Somebody else selling it already. This gets used to argue both ways, which is a good sign it argues nothing. A competitor proves people can be sold something in that shape. It does not prove there is room, and it does not prove the competitor is doing well.
• A waitlist with no price on it. Signing up for a free announcement is a time signal, and a weak one. Treating it as a sales forecast is where a lot of launches go wrong.
• Your own certainty. You are the one person guaranteed to want this product. That is why you thought of it.

What is the cheapest honest test?

Put the smallest real version on sale, at the price you actually want, in front of people who do not know you.

Each part is doing a job. Smallest real version means it has to be a thing somebody can use, not a promise. The price you actually want matters because a $3 test tells you nothing about a $49 product. People who do not know you is the part most tests skip.

The honest part matters more than the cheap part. If the product is not finished, say that in plain words on the page, give a delivery date, and refund anyone who changes their mind while they wait. Taking money for something you have not decided to build yet is not a test, it is a problem you are scheduling for later.

You need a page that takes a card, a file, and somewhere for the money to land. That is the entire setup. A link in bio with a checkout on it covers it for most people, and you can decide whether you need a company behind it once something has sold.

What does running that test cost?

Two costs, and only one of them is obvious.

The obvious one is the fee on each sale. On Stakk a $19 test product carries 10% + $0.50, which is $2.40, and card processing is inside that number rather than added after it. No monthly fee, so a month where nothing sells costs nothing.

The less obvious cost is the one charged before anyone buys. On a marketplace, listing can carry its own fee: Etsy's own fees page puts it at $0.20 per listing, charged regardless of resulting sales, with the listing expiring after four months. That is fine when something sells. It is a different shape of risk when you are testing ten ideas to see which one moves, because nine of them bill you for the privilege of being wrong.

Neither number is large. The reason to know both is that they push you toward different testing habits: per-sale pricing rewards putting up a lot of small tests, per-listing pricing rewards putting up fewer and better ones.

What do you do if nobody buys?

Resist the urge to conclude anything about the idea, because a silent launch has at least three explanations and they need different fixes.

Nobody saw it. They saw it and did not understand what it was. They understood it and did not want it at that price. Those look identical from the outside, and you cannot tell them apart without knowing how many people actually reached the page.

So check traffic before you rewrite the product. If a handful of people landed on the page, you have not run a test yet, you have run a distribution problem. If a few hundred landed and none of them bought, the page or the offer is doing something wrong, and the fastest way to find out is to ask three of them directly.

Change one thing at a time, and decide the rule before you start: how many visitors, by when, and what you will do at that point. Written in advance, that sentence is a decision. Written afterwards, it is a rationalization, and you will keep the product alive for months on the strength of it.

When is Stakk the wrong choice for this test?

When the per-sale cut is the thing you are trying to avoid. Systeme.io has a permanently free plan with no transaction fee at all, and Whop charges 2.7% + $0.30 with no monthly cost, so both of them cost less per sale than we do. If your whole test is "will one person pay", either of those is a reasonable place to run it and I would not argue with you.

Stakk is also simply unable to host some tests. There is no subscription billing, so you cannot test a membership. There is no course hosting and no call booking. Prices run from $1 to $999 and USD only. If your idea needs any of those, go elsewhere, and the platforms that hold your money often carry more of that surface. Our breakdown of who holds your money and who pays you directly is the fastest way to see which camp each one sits in.

What we do have, once the test is running, is the part that comes after the first yes: an order bump and a post-purchase upsell on every account, nothing behind a tier. The bump rides the same charge as the product, so it carries one fee rather than two. The upsell is a second charge and takes the full fee again, which is worth knowing before you build a two-step funnel on it.

Want to run the test this week?

You can have a real product page with a real checkout live in about ten minutes on Stakk. It is free to start, there is no monthly fee, and we take 10% + $0.50 a sale with card processing included. Your buyer pays into your own Stripe account, so the money is yours from the moment it clears rather than sitting in a balance we control.

That is the whole point of the test. Put one small thing at a real price in front of strangers and let them answer the question, because nothing else on this page can.

Start selling on Stakk, free