How to choose between startup ideas when you have several

Why having several ideas is a real problem

Having five ideas feels like abundance. In practice it is usually a way of not committing to any of them. Each idea stays at its most attractive while it remains hypothetical, and the moment one runs into difficulty there is always a fresher one to switch to. Months pass, nothing ships, and the list grows.

The goal of this process is not to find the perfect idea. It is to make a decision you can defend, spend a fixed amount of time testing it, and then either commit or move to the next one with a clear reason. A good process makes it cheap to be wrong and expensive to dither.

Step 1: cut the list to three to five

Comparing twelve ideas properly is too much work, so people compare them badly. Before any scoring, remove anything that fails an obvious filter. These are pass/fail questions, not ratings.

• Can you name a specific buyer? — Not 'small businesses' but a role in a type of organisation. If you cannot, the idea is not ready to compare.
• Is anyone already paying to solve this, even badly? — Money spent on spreadsheets, agencies or clunky software counts. No spending at all is a warning, not an opening.
• Could you build a usable first version in roughly three months with what you have? — If the honest answer needs a team or funding you do not have, park it.
• Would you still want to work on it after two years of slow progress? — If the answer is a shrug, it will not survive the first bad quarter.

Step 2: build a weighted decision matrix

A weighted decision matrix scores each idea against the same criteria, multiplies each score by how much that criterion matters to you, and adds the results. The crucial discipline is choosing the criteria and weights before you look at the ideas, so you cannot quietly tune the weights to favour the one you already like.

Keep it to five or six criteria that do not overlap. Typical ones are evidence of demand, your ability to reach buyers, feasibility of a first version, willingness to pay, and your own fit with the market. Weights should reflect your situation: a solo founder with no savings should weight speed to first revenue heavily; a funded team can weight market size more.

The table below is a worked example with made-up scores, using a 1-5 scale and weights that add up to 100 per cent.

Reading the matrix honestly

In the example, Idea B edges Idea A by a small margin, and Idea C falls clearly behind. The right reading is not 'build B'. It is 'C is out, and A and B are close enough that the scores cannot separate them'. A difference of a few tenths on a five-point scale is well within the noise of your own guesswork.

Look at where each idea is weak, not just at the total. Idea A scores 2 on reaching buyers: that is a single point of failure that could sink it regardless of demand. Idea B scores 2 on willingness to pay, which might be fixed with a different buyer or pricing model. Weaknesses you can test cheaply are less worrying than weaknesses that need luck.

Finally, run a sensitivity check. Shift the heaviest weight by ten points in each direction and recalculate. If the winner changes, your decision depends on a judgement call you should make explicitly rather than hide inside a spreadsheet.

Step 3: time-boxed tests with kill criteria written first

Take the top two ideas and give each a short, fixed test, typically one to two weeks. The point is to replace scores with evidence. A test might be ten conversations with the named buyer, a landing page with a real price, or an attempt to get three pre-orders or signed letters of intent.

Before you start, write down what result would kill the idea. For example: 'If fewer than four of ten buyers describe the problem without prompting, Idea A is out.' Or: 'If nobody agrees to a paid pilot after twelve conversations, Idea B is out.' Writing this first matters, because after the test you will be tempted to reinterpret weak results as promising.

Kill criteria should be specific, measurable and slightly uncomfortable. If you cannot imagine failing them, they are not testing anything.

Avoiding sunk cost and other traps

• Sunk cost — The time you already spent on an idea is gone whichever you pick. Ask only which idea is most promising from today. A domain name and a logo are not evidence.
• Novelty bias — The newest idea always looks best because it has not met reality yet. Score it with the same scrutiny as the old ones.
• Scoring your favourite last — Rate the idea you care most about first, before you have calibrated generously on the others.
• Endless re-scoring — If you keep adjusting scores without new information, you are procrastinating. New scores need new evidence.
• Parallel half-effort — Running three ideas at a third of your attention each usually produces three inconclusive results. Test two at most, then commit to one.

Making the final call

After the tests, you will typically land in one of three places. One idea passed and the other failed: commit to the survivor. Both passed: pick the one with the cheaper path to first revenue, because early revenue buys time to learn. Both failed: that is a successful result, since you have saved months, and you return to the list with sharper criteria.

Once you commit, set a review date roughly three months out with its own success measure. Until then, park the other ideas in a document and stop reopening them. A decision you revisit weekly is not a decision.

Questions