To evaluate a startup idea, score it separately on ten dimensions (problem, unique value, business model, timing, competition, feasibility, pivot potential, market size, go-to-market and team fit), back each score with evidence rather than opinion, and act on the weakest one first. One overall verdict hides the part most likely to kill the idea.
Most founders evaluate an idea by asking whether it is good. That question has no useful answer. Ideas fail on one specific weakness, and an overall impression lets a strong problem cover for a missing distribution channel until you have spent six months building.
The useful reframe is that evaluation is a search for the cheapest available evidence that you are wrong, organised so you can see where you are weakest. If you cannot find that evidence, you have learned something real. If you can, you have saved yourself six months.
This guide is the scoring framework Truewick is built around, written so you can run it by hand whether or not you use the tool. If you have not gathered any evidence yet, start with a 7-day validation plan, then come back here to judge what you found.
The ten dimensions to score
A single “is this good?” verdict is not actionable. What you need to know is which part is weakest, because that is what you fix or de-risk first. These are the dimensions Truewick scores, and they work equally well as a manual checklist. Score each out of 10. For what a strong and a weak answer look like on every one, see is my startup idea good?
| Dimension | The question it answers |
|---|---|
| Problem validity | Is this a real, documented pain — or an assumed one? |
| Unique value | Why you rather than the obvious alternative? |
| Business model | Does the money mechanism actually work at this price? |
| Timing | Why is now different from three years ago? |
| Competition | How crowded, and where is the genuine gap? |
| Technical feasibility | Can this team build it in a sane timeframe? |
| Pivot potential | If the wedge fails, is there a second shot? |
| Market size | Is the bottom-up number big enough to matter? |
| Go-to-market | Do you have a channel, or a hope of one? |
| Team fit | Why is this the team that wins this? |
Weakest dimension first. A 7/10 average hiding a 3/10 on go-to-market is not a 7/10 idea.
What the output looks like
Illustrative sample — not a real customer result
An AI scheduling assistant for independent physiotherapy clinics that books, reschedules and chases no-shows over SMS.
Better than 74% of ideas scored on Truewick
- 9/10Problem validityNo-shows are a well-documented revenue leak in small clinics; complaints are easy to find unprompted.
- 8/10TimingSMS automation plus cheap LLM call handling only became viable in the last two years.
- 7/10Business modelPer-clinic monthly pricing is proven in adjacent tools; margin depends on call volume.
- 6/10Unique valueDifferentiation rests on the no-show chase, not on scheduling itself.
- 5/10CompetitionCrowded with generic practice-management suites; the wedge is narrow but real.
- 4/10Go-to-marketWeakest dimension. Independent clinics are hard to reach and slow to switch; no obvious channel identified.
Read that scorecard the way an investor would: the average is respectable, but the decision is entirely about go-to-market. The right next move is not to build the product — it is to spend two weeks finding out whether you can reach twenty independent clinics at all. If you cannot, the other nine dimensions do not matter.
Where the evidence for each score comes from
1. Problem validity: find the complaint before you mention your product
Before you describe your solution to anyone, find people already complaining about the problem in public — support forums, Reddit threads, app-store reviews, G2 complaints, X replies. These complaints were written with no knowledge of you, which makes them the highest-quality signal available and the cheapest to collect. If you cannot find anyone describing this pain unprompted, that is your first real finding.
2. Competition and business model: check what people already pay
An existing paid alternative is good news, not bad. It proves budget exists. What you are looking for is the shape of the market: how many competitors, what they charge, and where their pricing pages reveal who they have chosen not to serve. A problem with zero paid solutions is usually a problem nobody is willing to pay to fix.
3. Market size: count customers from the bottom up
Ignore the top-down “$40bn market” framing — it is unfalsifiable and investors discount it instantly. Instead: how many businesses or people have this problem, what would they plausibly pay per year, and what share could you realistically reach in three years? A defensible small number beats an indefensible large one.
4. Problem and go-to-market: interview to disconfirm, not to confirm
Ten to fifteen conversations is usually enough to see a pattern. Ask about the last time the problem occurred, what they did about it, and what it cost them. Never ask whether they would use your product — that question produces the say-do gap on demand. If they have never taken any action about the problem, they will not buy a solution to it.
5. Business model: test willingness to pay before you build
The only reliable proxy for revenue is someone attempting to give you money. A landing page with real pricing and a checkout that collects payment details, a pre-order, a paid pilot, a deposit. Signups for a free waitlist measure curiosity, not demand, and founders routinely mistake one for the other.
6. Re-score, then write down what would change your mind
Update each dimension's score with what you found, and record (in writing, before you start building) the specific result that would make you stop. Founders who skip this step rarely stop, because by then the sunk cost is doing the reasoning for them.
What cited evidence looks like
Illustrative sample — paraphrased, not quoted from a real source
Scores are judgements. Evidence is checkable. This is the shape of what a search pass should return for the same idea — real complaints, each traceable to where it was said, written by people who had never heard of you.
“We lose about four grand a month to no-shows and I still ring people myself between patients.”
clinic owner, physiotherapy practice-management forum
“Our software sends one reminder 24 hours before and that's it. Nobody chases. It's the chasing that's the job.”
practice manager, allied-health subreddit
“Tried two booking tools, both assumed we had a receptionist. We don't.”
solo physiotherapist, app-store review of a scheduling product
The three failure modes to watch for in yourself
- Confirmation shopping. Asking ten friendly people and stopping when three are positive. Set the number of conversations and the disconfirming question before you start.
- Mistaking curiosity for demand. Waitlist signups, likes and “I'd definitely use that” cost the respondent nothing. Only a payment attempt costs something.
- Validating the solution instead of the problem. If your questions describe your product, you are testing your pitch, not the market. Ask about their last experience of the problem instead.
How long an evaluation takes
A first-pass score from desk research takes a couple of days: steps 1 to 3 need no one's time but yours. Backing every dimension with interview and payment evidence takes two to four weeks, dominated by step 4, because scheduling conversations is slow.
That compression is what Truewick automates: it scores the ten dimensions, benchmarks the result against more than 6,000 Y Combinator companies, and runs a cited web search for the complaints, pricing and demand signals behind steps 1 to 3 — so you arrive at your customer conversations already knowing what to ask. It does not replace step 4 or step 5. Nothing does.