AiFred

Okay let's begin. So this is the Q4 2025 retrospective. Thanks everyone for joining on this Thursday afternoon.

I'll start by sharing my screen with the OKR dashboard so we can all see where we landed.

First, what went really well. The new onboarding flow has been a massive win — we reduced time-to-productive for new engineers by 41%, which is actually better than the 35% target we set.

People are ramping up noticeably faster and the feedback from the last three hires has been very positive. Also, the design system adoption is now at 78% of UI components across both web and mobile.

That's huge progress from where we were six months ago. And finally, we passed the annual security audit with only four medium findings, no criticals or highs.

That's probably the cleanest result we've ever had.

On the other side — what didn't go well. The mobile app launch slipped by six weeks. We originally targeted mid-November and ended up going live December 23rd.

The crash rate on Android 14+ is still sitting at 2.1%, which is better than the 4% we had in beta but still too high for comfort.

A big part of the delay and the stability issues came from the repeated dependency blocks with the payments team — we lost almost the entire October sprint waiting on their side.

That was frustrating for everyone involved.

Another thing that came up repeatedly in the retro prep doc is documentation. Honestly, we're still in pretty bad shape here.

Half the backend services don't have up-to-date OpenAPI specs, some of the newer microservices have basically zero internal docs, and even when docs exist they're frequently outdated.

This is slowing down onboarding and increasing debug time across teams.

Also, meeting fatigue is real. We have too many recurring sync meetings — standups that could be async, planning sessions that run long.

Refinement that sometimes feels like group therapy instead of focused backlog work. Several people mentioned they feel they have almost no deep work time left in the day.

And probably the heaviest topic — morale has taken a hit. There were three separate re-org rumours between September and December.

Even though none of them materialized in the way people feared, the uncertainty and the lack of clear communication from leadership really wore people down.

A few people said they feel emotionally drained or even a bit cynical at this point, which is not where any of us want the team to be heading into 2026.

Okay, let's move to actions — what we can actually do differently.

First proposal: replace at least 60% of our recurring meetings with async video updates — Looms, short screen recordings, whatever works.

We keep the daily standup but make it optional async unless someone really needs to ask something live.

Second: we need a documentation ownership rota. Each service gets a clear owner responsible for keeping the README, API spec, and architecture decision records up to date.

We can start small — maybe one service per squad per quarter.

Third: let's do a retro on our retros. The format is getting a bit stale and we're spending too much time in ceremony.

Maybe experiment with shorter time-boxes or different facilitation styles.

Fourth: someone — probably leadership — needs to give a clear statement on organizational stability for 2026 within the next two weeks.

Even if it's just 'no re-org planned in H1', that would help a lot.

Fifth: trial a no-meeting Thursday every week for at least February and March. Protect that as focused engineering time.

See if it moves the needle on velocity and happiness.

And last one for today: revisit how we set OKRs next cycle. Fewer stretch goals, more realistic committed targets.

The all-stretch-all-the-time approach burned people out this quarter.

I'll summarise everything into a short doc with owners and target dates and send it by end of day tomorrow. Please comment or add anything I missed.

Thanks again everyone — let's make 2026 better than 2025. Have a good rest of the day.