How to decide what not to build
Everything on the list looks important. How do I cut it?
Rank by the evidence behind the request, not by how loudly it was made. Anything with no named buyer, no observed behaviour and no revenue attached goes to a not-now list with the reason written next to it. A backlog is a record of unmade decisions; the useful artefact is the list you refuse.
Take every open item and attach three facts: who asked, what evidence exists that it changes behaviour, and what it costs in founder-weeks. Items without all three do not get scheduled. That single rule usually removes half a backlog in one sitting.
| Signal | Weight |
|---|---|
| A paying customer stopped or nearly stopped because of it | Highest |
| Blocks money already agreed | High |
| Multiple named buyers asked, unprompted | Medium |
| Would be nice; a competitor has it | Not now |
| We are bored of the current thing | Not now |
Write the refusal, not just the ranking
For every not-now item, write one line: why not, and what would change the answer. Six weeks later that line stops the same argument from being had again, and occasionally tells you the condition has now been met.
Keep the refusals. A written not-now list is worth more than a prioritised backlog.
Do this next
- List every open build item with who asked and what evidence exists.45 minutes
- Move everything without evidence to a not-now list, with a reason line each.30 minutes
- Pick the two items that unblock money and schedule only those.15 minutes
Avoid
- —Ranking by effort estimate alone.
- —Building for the loudest customer instead of the representative one.
- —Keeping a 200-item backlog because deleting feels like losing work.
“From my current build list and customer records, what has real evidence behind it and what should go to not-now?”
Ask Aury thisLast edited 2026-08-18