Choosing between business ideas is settled by evidence far faster than by deliberation. Score each candidate on four predictors — how reachable the buyer is, whether they already pay to solve the problem, how many days until you get a signal, and whether you would still care in twelve months — then run the cheapest test first rather than choosing outright. Most founders can test two ideas inside a month, which converts an argument into two answers. The idea you commit to should be the one that survived a test, not the one that won a spreadsheet.
Force each idea into one structure: "[specific buyer] currently solves [problem] by [expensive workaround], and it costs them [time or money]."
This single exercise settles a surprising number of comparisons in ten minutes. One idea will have a named buyer and a costed workaround; the other will have "small businesses" and a hopeful assumption. That gap is the decision, and it is invisible while the ideas live in your head as vibes.
If an idea cannot be written in that shape at all, it is not yet an idea — it is a product concept looking for a problem. Park it and note what you would need to learn to fill in the blanks.
Two candidates is a false choice more often than a real one. Before committing to a comparison, spend twenty minutes generating more against your actual constraints — your skills, your reach, your starting budget.
The generator below returns four businesses matched to what you can genuinely put in, each with the specific catch and a first test. Add the good ones to your list. It is common to discover that the two ideas you were agonising over both score worse than something you had not considered, and that discovery costs you an afternoon rather than a year.
Aim for a pool of five or six candidates before you start scoring. Comparing two is a coin flip with extra steps.
Skills, a trade, an industry, a community you're part of. The more specific, the less generic the ideas.
Ask one question per idea: could you get fifty of these buyers onto a list this week, without an introduction?
Ideas fail on access far more often than on quality. A mediocre idea aimed at agency owners — who answer LinkedIn messages — beats an excellent one aimed at hospital procurement, who do not. For a solo founder with no network, reachability is close to destiny, and it is knowable in an afternoon.
Score it bluntly: yes, hard, or no. Anything scoring "no" leaves the list regardless of how good it looked, because you cannot run the tests that would tell you whether it was good.
Two more columns, both quick.
Existing spend: does the buyer already pay someone or something to handle this — an agency, a tool, a part-time employee, a subscription? Existing spend means a budget line exists and you are competing for it. No spend means you must create one, which doubles the sales cycle and is the most consistently underestimated risk on any list.
Time to evidence: how many days until this idea produces a signal you would act on? An idea testable with a landing page and forty messages in a week beats a more interesting one that needs a working prototype, because the fast idea either validates or frees you — and both are progress.
Score both columns high, medium or low and look at the pattern rather than a total. The candidate you want has existing spend and fast evidence; a candidate with existing spend but a three-month test is a real business you cannot afford to check yet, and one with fast evidence but no existing spend will produce a clear, cheap answer that nobody wants to pay for it.
Ask which candidate you would still be working on in month twelve, when it is grinding, unglamorous and slower than you expected.
This is the one subjective factor worth scoring, because most businesses die of founder attrition rather than market rejection. An idea you find boring on day one will not survive a year of quiet weeks, whatever the spreadsheet says.
Weight it, but do not let it win alone. The combination you are looking for is an idea that scores well on reachability and existing spend and that you can tolerate for a year. An idea you love that nobody can be reached to buy is a hobby with a business plan attached.
Whichever candidate scores highest, start with the test that costs the least time. If it validates, you have your answer. If it fails in a week, you have lost a week and got the other candidates back.
Write both the test and its pass mark before you begin: "fifteen deposits from 500 visitors" or "one signed pilot from forty messages". A threshold decided afterwards will always be reinterpreted generously.
Most founders can run two of these inside a month. That is the real advice hiding in this whole procedure — the choice you are agonising over is usually cheaper to settle empirically than to settle by thinking harder about it.
kitty.build takes each candidate and runs it through nine specialist boards with live research, returning a cited verdict and a comparable score for each — so the comparison is made on evidence rather than on which idea you described more enthusiastically.
Feed her an idea