Here's the situation I was trying to fix.
I had a list of eleven side-business ideas and no way to tell which ones deserved a weekend. Every idea sounded plausible on a Tuesday night. Every one felt equally worth pursuing. So I kept not starting any of them, which felt like indecision but was actually a lack of a filter.
The problem wasn't motivation. It was that "should I build this?" is an unanswerable question. It has no edges. There's always more research to do, another competitor to check, another reason it might not work.
So I built a test with edges. Two hours, on a Saturday, with a defined output. It doesn't tell you whether the business will succeed. It tells you whether it deserves a second weekend—which is the only decision in front of you right now.
This is the first piece in After Hours, On Purpose, the category about building a side path when you have a job, a family, and finite energy.
Why Most Side-Business Evaluation Fails
Three failure modes, and they're all about timing and scope.
Research Is a Substitute for Action
Reading about a market feels productive. It's not. It's the cheapest form of work, and it expands to fill whatever time you give it. Six weekends of research can teach you less than two hours of trying to make one actual thing.
The research trap is especially cruel because it feels responsible. You're being diligent. You're not rushing in. You're doing your homework. Meanwhile the idea you were excited about three weeks ago has quietly become a project you're avoiding.
You're Evaluating the Wrong Question
"Will this make money?" can't be answered before you build something. "Do I have a skill for this?" is too vague to be useful. "Is this a good market?" is a question about the market, not about you.
The only question you can answer in two hours is narrower: can I produce one small, honest piece of evidence that this might work?
Weekends Are Too Expensive to Waste
A Saturday morning is a real resource. If you have a job and a family, you have maybe two usable blocks of time a week, and one of them should probably go to rest. Every idea you spend a weekend on is an idea you didn't spend that weekend on instead.
This is why the test is two hours and not a weekend. Two hours is cheap enough to spend on a maybe. A whole weekend is not.
The Test
Four phases, two hours total, hard stop at the end.
Phase 1: Define the One Thing (15 minutes)
Before you touch anything, write down two sentences:
The offer: who it's for and what they get. Not a business plan. One sentence.
The evidence that would matter: what would you need to see, in two hours, to believe this deserves a second look?
The second sentence is the one people skip, and it's the one that makes the test work. You're not testing whether the idea is good. You're testing whether it produces a specific, pre-defined signal.
Examples of honest evidence:
One stranger describing the problem in their own words
One person saying they'd pay for a rough version
One draft of the deliverable that you'd actually be willing to send
One piece of proof that the thing you'd sell can be made in a reasonable time
If you can't name what evidence would matter, you're not ready to test. You're still deciding what the idea even is.
Phase 2: Produce the Artifact (60 minutes)
Now make the smallest real thing. Not a plan for the thing—the thing itself.
Depending on the idea, the artifact might be:
A one-page outline of the service, written as if you were sending it to a client
A rough version of the deliverable—the report, the template, the audit, the setup
A single outreach message you'd actually send, written and ready
A version of the workflow for a hypothetical client, run once on real or realistic data
The rule: it has to be something a real person could react to. Not a concept, not a plan, not a slide. A thing.
Set a timer. Sixty minutes. When it's done, stop, even if it's rough. Rough is expected—this is a test, not a launch.
Phase 3: Test It Against One Real Person (30 minutes)
This is the phase most people skip, and it's the one that turns an idea into information.
Find one person who fits the audience—a colleague, a former coworker, someone in a community you're part of. Not a friend who'll be encouraging. Someone who'd actually have the problem.
Show them the artifact. Ask three questions:
Does this describe a problem you have? (Not "is this interesting?"—does the problem exist?)
If it existed, would you use it? (Not "would you pay for it?"—that comes later, and people lie about hypothetical purchases. Start with use.)
What's wrong with it? (This is the most valuable question. The answer tells you what you misunderstood.)
Take notes. Don't defend the artifact. Don't explain what you meant. Just listen. Thirty minutes, one person.
Phase 4: Decide (15 minutes)
Write down three things:
What I learned. One or two sentences. Not "it went well"—what did you actually learn?
The signal. Did the evidence you defined in Phase 1 show up? Yes, no, or partially?
The call. One of three:
Continue: the signal showed up, and the next test is obvious.
Adapt: the problem is real but your version missed—adjust the offer and test again next Saturday.
Stop: no signal, or the problem doesn't seem real, or the artifact was painful to make in a way that suggests the work isn't for you.
All three are valid outcomes. Stop is not failure. Stop is the test working—it just saved you a month.
The Before and After
Two real examples from my own list.
Idea A: A weekly reporting package for small businesses.
Offer: A monthly reporting setup for small agencies that don't have an analyst.
Evidence that would matter: One agency owner describing the reporting problem in their own words.
Artifact: A one-page outline of the package, written as a client-facing description.
Test: Showed it to a former colleague who runs a five-person agency.
Result: She described the problem immediately—"I do this every Sunday night and I hate it"—and pointed out that she'd want it done for her, not handed to her as a template.
Call: Adapt. The offer was right; the delivery model was wrong.
Idea B: A paid workshop on AI workflows for job seekers.
Offer: A two-hour workshop for professionals using AI in a job search.
Evidence that would matter: One person saying they'd pay for a rough version.
Artifact: A rough outline of the workshop, one page.
Test: Showed it to two people in a career-transition Slack community.
Result: Both said it sounded useful. Neither could name a specific problem it solved. "Useful" is not a signal.
Call: Stop. The topic is interesting; the offer isn't specific enough to be worth building.
Two hours each. One continued, one stopped. The stopped one would have taken a month to figure out the slow way.
What Still Needs You
The parts of this test that can't be automated or outsourced:
Defining the evidence. Only you know what signal would actually change your mind. A model can suggest options, but the choice is a judgment about your own risk tolerance and goals.
Making the artifact. The point is to find out whether making the thing is something you'd want to do repeatedly. If the sixty minutes felt like torture, that's information. A model can't feel that for you.
Choosing the person to test with. One real person in the audience beats ten encouraging friends. Picking them is a judgment about your network and who'll be honest.
Reading the signal. People are polite. "That's interesting" is not a yes. "I would use this" is warmer. "I would pay for this" is the strongest signal, and still not a promise. Interpreting the gap is human work.
Making the call. Continue, adapt, or stop. The model can lay out the options; it can't decide for you, and it shouldn't. The consequences are yours.
The Two-Hour Version
If you have an idea and a Saturday morning:
15 min: Write the offer sentence and the evidence that would matter.
60 min: Make the smallest real artifact. Timer on. Stop when it rings.
30 min: Show it to one real person in the audience. Ask three questions. Listen.
15 min: Write what you learned, the signal, and the call.
Two hours. One of three outcomes. And on Sunday, you'll know whether next Saturday belongs to this idea or a different one.
The point isn't to build a business in two hours. It's to stop spending weeks on ideas that would have revealed themselves as dead ends in a single morning.
Better work first. More options next.
Make the workflow earn its place.
No feedback yet — submit the first.