A filter for what to build, now that building is cheap
Since the summer I have built an appointments platform for the Romanian market, a QR based feedback tool for restaurants, an online shop, a search product and something that reads public tender documents and tells you which are worth bidding on.
They all work. Each took between a week and a month. One of them has users who are not me, and working out why that one is different gave me a filter I now run on everything.
What changed, mechanically
For fifteen years "should I build this" and "can I afford to build this" had the same answer, because building was expensive. If an idea was not clearly worth a quarter, it did not get built. That filter did a lot of quiet work on my behalf and I never had to think about it.
A weekend now buys what a quarter used to. So an idea no longer has to be good to survive. It only has to be interesting on a Saturday morning.
Nobody warns you about this because from the inside it feels like productivity.
The filter
Before anything gets started:
Name the person. Not a segment, a person, who I can message when it is ready. "Restaurant owners" is not a person. If I cannot name one, the project is a hobby, which is fine, but it goes in the hobby budget and gets a weekend, not a month.
Say what they do today instead. If the honest answer is "nothing, this is not a problem they have", I have invented a customer. If the answer is "a spreadsheet", good, spreadsheets are the strongest competitor in the world and beating one is a real achievement.
Decide the distribution before the first commit. How does a second user arrive. Write one sentence. If the sentence is "I will post about it", that is not distribution, that is hope.
Set a kill date. A date, in the calendar, by which it either has that person using it or gets archived. Archived honestly, not left in a folder pretending to be an asset.
What the one that worked did differently
The appointments platform has people booking through it who do not know me.
It got there because I was forced into the unglamorous things: proper Romanian copy, the timezone handled correctly, notification emails that do not look like spam, the legal pages the payment processor demands before it will let you charge anyone, and then going and telling people it exists.
None of that was hard. All of it was boring. Boring is the part where speed helps you not at all. Nobody has made distribution ten times faster, and that is now the entire bottleneck of a small software business.
The honest arithmetic
Four months bought me five working products and one business.
With the old filter I would have built one thing and it would probably have been the wrong one, because I would have had to choose in advance without evidence. So the speed is not wasted. It is a cheaper way to be wrong.
But only if you let the wrong ones die. Five living projects with no users is not a portfolio, it is five sources of guilt and five sets of dependencies to update. The kill date is the part that makes the rest of it work.