In short: the product is proven and the number still fell, so the problem has moved to distribution; stop building, name one story, and work one channel.
Key points
- A month of releases with a falling number usually means the product objection is dead and there is no lead story.
- Shipping is the comfortable work because releases are visible and finite; distribution is slow and repetitive.
- Freeze new features for sixty days in writing, while fixes and security work continue.
- Report the core number with and without incentives and free tier every week.
- It worked if the number without incentives rises and a partner brings buyers without being chased.
What it usually means
The call comes after a big month of releases. The team moved fast. The core number, deposits or weekly active teams or paid seats, went down anyway.
Three things are usually true at once.
First, the product objection is dead. The thing works. It has worked for real users, repeatedly, across versions. Nobody who churned said it was broken.
Second, there is no lead story. The team has been telling three stories to three audiences. Each one is plausible. Together they cancel out. A buyer who hears three reasons to care hears none.
Third, shipping was the comfortable work. A release is visible, finite and hard to criticise. Distribution is slow, repetitive and easy to criticise. Given the choice, most teams build.
An onchain protocol launches a dozen new markets in a month and deposits fall. An AI product adds an integration every week while weekly active teams stay flat. The pattern is the same. Supply went up. Demand did not follow, because nobody was working demand.
This is a good month, even if it does not feel like one. It removes an explanation. The team can no longer say the product is not ready.
Where it traces to
This traces to Part 1 and Part 8.
Part 1 asked for one system, not three problems. One named buyer, one calendar, one owner. A team running three stories has skipped this. The shipping cadence filled the gap where the buyer definition should have been.
Part 8 covered distribution when platforms are closed. The channel sort, open, restricted, closed, was meant to be done before the roadmap was set. If it was never done, releases went out into no channel at all.
In crypto this also traces to Parts 19 and 20. New markets without issuers, integrations and curators lined up are supply with no demand. Protocol revenue does not follow from more venues. It follows from the venues that already have the depositor.
In AI it traces to Part 13. A shallow integration is not distribution. The partner who already has the team is.
The first move this week
Stop building for sixty days. Fixes and security work continue. New features do not. Put it in writing so the team knows it is a decision, not a mood.
Name one lead story. Not the best one, the one you can prove. Write it in a sentence a partner could repeat without you in the room.
Pick one channel band and work it exclusively. For most teams at this point it is partners who already have your buyer. They have the trust and the list. You have the product that works. Make a list of twenty, contact all of them, and run it like Part 4 said to run a round.
Report the core number two ways from this week on.
| Line | What it shows |
|---|---|
| With incentives and free tier | What the market will take when paid to take it |
| Without incentives and free tier | What the market will pay for |
If the second line is flat while the first moves, you have been buying the number. Now everyone can see it.
What not to do
Do not ship one more feature. The next one will not be the one.
Do not announce more partnerships. A logo on a slide is not a partner working your buyer.
Do not blame the market. The market did not change in the month you shipped twelve things. Your attention did.
How you know it worked
Two signs. The core number without incentives moves up, even slightly, within the sixty days. And at least one partner is bringing buyers without being chased each week.
The checklist
- [ ] A written sixty-day build freeze, with the exceptions named
- [ ] One lead story in one sentence, and one channel band with a list of twenty
- [ ] The core number reported with and without incentives, every week
Common questions
Why did our metrics drop after a big product release?
Usually because the product works but there is no single lead story and nobody was working demand. Supply went up and demand did not follow.
What should a startup do when shipping more features doesn't grow usage?
Stop building new features for sixty days, name one lead story you can prove, and work one channel band, often partners who already have your buyer, starting with a list of twenty.
How can I tell if incentives are inflating my growth numbers?
Report the core number with and without incentives and free tier. If the second line is flat while the first moves, you have been buying the number.
Do more integrations help an AI product grow?
A shallow integration is not distribution. The partner who already has the team is.