I turn down more work than I take. Not out of principle — out of arithmetic. A project that fails costs me four weeks and a reference; a project I refuse costs me an email.
Six categories, roughly in order of how often they appear.
01"I have this idea, I just need someone to build it"
No users, no validation, no plan for the first ten customers. This is not a software problem yet; it is a demand problem. Building it produces a very expensive way to find that out.
The sentence:
"Before building anything, I'd want to see ten people try to buy this
by hand. If you can get ten, I'll build the thing that makes it
easier. If you can't, software won't fix that."02"Just like X, but cheaper"
X has forty engineers and three years of iteration. The request is not a build, it is a race with a much better-funded opponent.
The sentence:
"We can't beat them at the whole product. We could beat them at one
workflow they handle badly — which one do your customers complain
about?"03The owner is not in the room
You are talking to a marketing manager who is talking to a founder who has not decided anything. Every requirement is provisional, every decision needs escalation, and the project will drift for months without ever being cancelled.
The sentence:
"Can we get the person who can approve the budget on the first call?
I've never had a project go well where the decision-maker joined
at the end."04"Can you also do the marketing"
They want a partner who will make it succeed, not a builder who will make it exist. Those are different jobs, and pretending otherwise sets up a failure where you are blamed for the outcome.
The sentence:
"I can build the thing and make sure it works. I can't promise it
sells — nobody can, and anyone who does is guessing."05The budget that cannot buy the smallest useful version
Sometimes the idea is fine and the money is not. A $600 budget for a $3,000 outcome is not a negotiation, it is a mismatch.
The sentence:
"At $600, the honest version is one page with a form — no accounts,
no dashboard. That might be enough to start. It won't be the whole
thing."And if even that is too much: "come back when it's paying for itself". About one in five does.
06Anything I would have to lie about
Regulated categories where the client has no licence, platforms whose terms forbid the use case, features sold as compliant that are not. This is the only category where I do not negotiate at all.
The sentence:
"I can build it. You won't get it approved without the licence.
I'd rather tell you now than after you've paid."07Why refusing sells
A stranger saying yes to everything is telling you nothing — they need the money, so the yes is not information. A stranger who says "this part I won't do, and here's why" has just demonstrated that their judgement has a floor.
Every refusal I have sent has produced one of three outcomes: they narrowed the scope and hired me, they went away and came back later, or they told someone else that I was straight with them. All three are better than four weeks on a project that was going to fail.
Worth keeping
- Refuse when the problem is demand, not software.
- Escalate to the decision-maker before quoting, not after.
- A refusal with a reason is the cheapest credibility you can buy.