unless someone pays, don't code

Two students showed me an investment research tool. 200 friends on campus using it for free. Two brokerages paying per report. The plan for the next three months was to expand from 1,000 companies to 5,000 and build value chain maps.

I told them to stop. Your 200 friends aren’t paying, so they’re not users. The brokerages are paying per report, so you’re a dev shop. I can get that on Fiverr. Unless you can get one person to pay a recurring subscription, you should not write another line of code.

This is the thing I say most to people before they build. People say nice things when nothing is at stake. Everyone wants to encourage founders. The moment you ask “will you pay for this”, they say build one more feature, get me more data, come back next quarter. That’s the only real feedback you’ll ever get, and most founders avoid it for a year by coding.

Sell before you build. Find someone who already spends money on the problem and ask what they’d pay if you solved it. Get them to say a number out loud. “If you cut my token bill by half, I’d pay 5k a month”. That’s not a contract. It’s still better than anything else you have. If nobody will say a number, that’s market feedback. The product is wrong, or the customer is.

Apple showed iPhones on stage that didn’t work. Musk’s robots on YouTube are demos. Selling happens before the product exists. You’re smart enough to build it. Nobody doubts that. Selling is the hard part, so it goes first.

And charge from day one. Even 100 rupees a month. Free users will always say it’s good because it’s free. A founder was giving his chemistry tutor to 90 students for nothing and calling it validation. It’s not. Put a paid plan on it, release it to your followers, give them a day’s trial. If it has bugs, no one cares at 100 rupees. Collect the emails and give them three months free as apology. Now you have real validation and a list.

Impact is a word people use to avoid this. Impact is the result of your company making money. Without dollars there is no impact, there’s a Wikipedia banner asking for donations every year.

the call where i said it best

A student from a manufacturing family wanted to solve “fragmentation in indian manufacturing”. I asked what one factory would pay her for this week. She didn’t know. So we found one: the e-way bill. Every truck dispatch, same day filing, the owner does it himself on a Friday evening because it doesn’t work on mobile. Automate that and a factory pays 5 to 10k a year happily. That’s a foot in the door. The fragmentation thesis can wait until twenty doors are open.

do this this week

Go to the one person most likely to pay you and ask: if this worked, what would you pay, and when. Write the number down.

the line to remember

unless you can get a paid user for this, you should not code at all.