Our Q3 roadmap had 9 items. We shipped 2. A calm post-mortem

We are a 30 person health tech startup in Stockholm. In June I presented a Q3 roadmap with 9 items. Leadership nodded. Engineering nodded slightly less. On 30 September we had shipped 2. Here is what happened, without drama, because I think it is useful and because I promised my team I would write it down.

The 9 items, and what happened to them

1. Patient reminders v2: shipped.
2. Clinic dashboard export: shipped.
3. New clinic onboarding: 60% done, paused.
4. Integration with a second journal system: blocked, waiting for the vendor's API access since July.
5. Medical device documentation update: not a feature, but it took 6 weeks of two engineers. Part of our software is regulated as a medical device, so a change in intended use means documentation. Nobody had put it on the roadmap, because nobody thinks of paperwork as work.
6. Self-serve billing: deprioritised in August for a sales promise.
7. The sales promise: a custom report for one large region. Not on the roadmap. Shipped, actually. So we shipped 3 things, but only 2 of the 9.
8. AI-assisted triage notes: paused for a legal review of how the EU AI Act applies to us.
9. Performance work: half done, the way performance work always is.

What I would do differently

• Plan for 60% capacity. We planned for 90%. Support, on-call, sick children, vacations (it is Sweden, July does not exist) and regulatory work ate the rest.
• Put compliance on the roadmap. If it takes engineers, it is roadmap work. Hiding it makes the roadmap a lie and makes the engineers look slow.
• Now, Next, Later instead of quarters. Quarters suggest dates. Dates become promises. Promises become sales decks.
• One person says no to sales. Me. I did not, in August. I should have, or at least made the trade explicit in writing: "the custom report means self-serve billing moves to Q4". I said it in a meeting. I should have written it down.
• External dependencies get a give-up date. If the vendor API is not there by a set date, we stop waiting and replan.

What went right

The two things we shipped are used by clinics every day. The team did not burn out. Nobody worked weekends. That is lagom, and I will take it over 9 half-shipped features.

Roadmaps are hypotheses, not promises. I believe it more after this quarter. Happy to be proven wrong!