Where I Stand
- WhatsApp: 30 engineers, 450M users. FeatureOS: small team, $15K MRR, unclear how many active users. The ratio tells the story — WhatsApp achieved absurd leverage by saying no to everything except reliable messaging. What is FeatureOS's equivalent of "reliable messaging"?
- Jan Koum's default was "no." My default is "maybe later" or "let's explore that." These sound diplomatic. They're actually worse than "no" — they keep options open without closing anything, which means the team builds for ambiguity.
- WhatsApp had no code reviews, no scrum, no managers — because the people WERE the process. That only works when every person is elite and trusted. Do I have that level of trust in every person on my team? If not, I need either better people or better process. Not the limbo I'm probably in.
- $1/year covered everything. The business model was so simple it was invisible. FeatureOS has tiers, add-ons, usage limits, feature gates. Is the pricing complexity creating overhead that a simpler model would eliminate?
- WhatsApp dogfooded aggressively — every build was used internally before release. Am I using FeatureOS daily as a real user, or am I only in the admin view?
- Jean's career advice: focus on foundations, not frameworks. Am I building foundational product capabilities in FeatureOS, or chasing framework-level features that won't matter in 3 years?
What I Need to Change
- List every feature in FeatureOS. Mark the ones that fewer than 10% of customers use. Consider killing them. WhatsApp's power came from the features it DIDN'T build.
- Establish a clear "no" policy. When a feature request comes in, the default answer is no unless it directly reduces churn or increases activation. Write this down and share it with the team.
- Simplify pricing. Can FeatureOS be explained in one sentence, with one price? If not, the model is too complex for the stage.
- Start dogfooding FeatureOS weekly with the full team. Use it the way a customer PM would. If something feels off, it doesn't ship.
The Hard Question
If you could only keep five features in FeatureOS and had to delete everything else, which five would you keep — and why haven't you already made those five perfect?