The reversible bet, worked example: Testing a product idea with £500 and a deadline

The reason most professionals never test an income idea is not fear of failure. It is fear of an open-ended commitment: the sense that starting means months of evenings, thousands of pounds and a decision they cannot take back. The reversible bet is the answer to that fear, and an earlier post on this site set out the principle. This one sets out the practice: a four-week test of a physical product idea with a £500 ceiling and a fixed end date, worked line by line so that you can copy the structure for your own idea.

It uses a product because that is the path I know best and because products are the case where the fear of commitment is most rational. The same structure works for a consulting offer or a digital product with smaller numbers, and the last section shows how. Everything here is illustrative; the figures are the kind of figures such a test produces, not a promise about yours.

What makes a bet reversible

The cost ceiling is the amount you can lose without it changing anything at home. For most employed professionals that is somewhere between £200 and £1,000; this example uses £500. The end date is the point at which you stop and decide, regardless of how it feels; four weeks is long enough to get real answers and short enough to fit around a job. The question is the thing you do not know and cannot find out by thinking harder. For a product, the question is almost always the same one: will strangers pay a price that leaves a margin?

The four weeks, line by line

The evenings it actually takes

Money is the ceiling people worry about; time is the one that actually breaks tests. A four-week product test around a full-time job needs roughly ninety minutes on four evenings a week, and it needs those evenings to be assigned in advance rather than found. Week one is spreadsheet and supplier emails. Week two is photographs, the listing and the first parcels. Weeks three and four are three short videos each, replies to questions, and packing whatever sells. None of it requires a weekend, and none of it should be allowed to take one, because a test that eats the family's Saturday will be abandoned in week three regardless of the numbers.

Write the time ceiling next to the cash ceiling. Twenty-four hours over four weeks is a reasonable figure for this example, and if you find yourself at thirty, that is data too: either the product is more work than the margin justifies, or you are polishing when you should be counting. The job comes first during the day without exception. A parallel income that costs you your salary is not parallel, and a test that damages your performance at work has failed even if every unit sells.

Writing the criteria before you start

The step people skip is the one that makes the bet reversible: writing down, in week one, what result would mean "continue", what would mean "change one thing", and what would mean "stop". Without it, week four becomes an argument with yourself in which the sunk cost always wins. The criteria do not need to be sophisticated. They need to be written before you have any data to bias them.

What a test cannot tell you

A four-week, £500 test answers one question: will strangers pay. It does not tell you whether the product can be sourced reliably at scale, whether a competitor will undercut you in March, or whether you will still want to pack parcels in month nine. Those are questions for the next bet, and the mistake is to try to answer them all at once with the first one. Each bet should be sized to the question it answers and no larger.

It also cannot tell you much about a product that depends on paid traffic to sell at all. Some products only work with advertising behind them, and a test that excludes advertising will return a "stop" for those. That is a limitation worth knowing, but for a first bet with a job and £500 it is the right limitation: a product that cannot find its first dozen buyers without paid reach is not a product a part-time operator should be starting with.

The other thing a test cannot do is make the decision for you. It produces a number; you wrote the criteria; the two either match or they do not. People who run a good test and then ignore the "stop" result because they have grown fond of the idea have wasted the £500. People who ignore a "continue" result because it did not feel like success, since twelve sales is not a business, have wasted it in the other direction. Twelve sales at full price to strangers in four weeks, with no advertising, is exactly what success looks like at this stage.

The same structure on the other paths

A consulting offer tests with almost no cash. The ceiling is time: ten evenings. The question is whether one person will pay a fixed fee for a fixed piece of work. The test is three emails to people who already know what you do, one scoped offer, and a date by which you either have a paid engagement or you do not. The first consulting client post gives the offer, the price and the email.

A digital product tests with a pre-sale. The ceiling is a weekend to write the outline and a sales page; the question is whether anyone will pay before the thing exists. Ten pre-orders at a modest price is a continue; two is a change one variable; none is a stop, and the outline cost you nothing but the weekend. Content tests last, because its feedback loop is slowest, and its version of the bet is thirty posts on one platform in thirty days with the numbers written down at the end.

Why I still use it

My first manufacturer order in December 2024 was larger than £500, because by then I had spent six months on the question and was betting on a supplier rather than an idea. But it was sized the same way: to an amount I had written down as the most I could lose without my family noticing. The business that followed passed £500,000 in its first twelve months, and every expansion since has been a bet with a ceiling and a date. The size of the bets changed. The structure never did.

That is the method. Not a leap, and not a plan so detailed it never starts. A bounded, dated question, answered by strangers with money, followed by the next one. Four weeks and £500 is enough to find out whether your idea deserves a second bet. It is also enough to find out that it does not, which is the cheaper of the two lessons and the one most people pay far more to learn.

This is general information, not financial or business advice. All figures in the worked example are illustrative; your costs, fees, sales and results will differ, no income is promised, and income from selling goods or services may be taxable and reportable to HMRC.

Build first. Leave second. Choose third. Start with the assessment →