I ran a tech debt workshop to talk through what we’re going to do in the last 4 months of a year, given a 20 or 35% tech debt fixing budget. Tech debt directly impacts Accelerate metrics (deployment frequency, lead time). See also batch size for why smaller debt fixes are preferable.
Note to self:
- Have a parking lot for important, but off-topic items.
- Have a scribe.
Rough notes follow.
- Intro w/ most painful tech debt on your team
- Expectations of the workshop
- Would like to exit this with an understanding of things we can/should/will work on in 2025 (n=3-5).
- Would like to get you startd on your 2026 list.
- Shared understanding of how we can think through these debt tradeoffs
- Highlight any cross-team dependencies
- Communicate up and out what our technical debt fixes are and why.
- Generally foster a culture of talking about this stuff (debt)
- Let’s get stoked about making things better.
- Reminder: Holidays are coming up.
- Answer the question: What technical debt are we fixing in Q4? Why does it matter and what does it unblock?
- Talk through the specifics of 20-35% debt fixing budget
- Align on the meaning of tech debt
- maintenance vs refactoring vs rearchitecting
- Align on the expectation that all EMs maintain a backlog of technical issues (including debt)
- Reiterate that while there may be some externally imposed technical debt items, this should also come from them.
- Categorization framework
- Separate into groups, brainstorm debt items (focus on impact; can be multiple kinds of impact)
- Consolidate ^^ into 5-7 items per group
- Impact Estimation
- Work through 2-3 examples together
- Quicker version: risk/friction matrix (blastfrequency) + (velocitytrend)
- Blast radius: (1=single team, 3 = critical path for product area, 5 = company-wide issue)
- Frequency of pain (1 = quarterly, 3 = weekly, 5 = daily)
- Engineering velocity (1=slight slowdown for some stuff, 3=blocks certain types of features, 5=grinds multiple teams to a halt)
- Trend (1=stable annoyance, 3=slowly degrading, 5= accelerating problem)
- Score 30+: Critical - Q4 candidate, Score 20-30: Important - Q1 2025, Score 10-20: Moderate - roadmap it, Score <10: Monitor
- Weekly Engineering Tax = Direct Time (they spent x time) ~~ Indirect Time (things take longer b/c…) ~~ Opportunity Cost (can’t build x, because..)
- eventually turn ^^ into engineering months then converted to dollars.
- Team scoring. Groups estimate/assess impact of their items
- Prioritization
- effort vs impact
- quick vs strategic vs someday
- confidence high/medium/low in ability to execute
- surface any organizational/political challenges
- Dependency mapping: What would block you? Include missing skills/capabilities/cross-team issues
Possible outcome for each item
Item: [Description] Confidence: Green/Yellow/Red Team Owner: [Name] Estimated Effort: [Days/Weeks] Skills Gap: [None/Specific skill needed] Research Needed: [None/1 sprint/2+ sprints] Q4 Viability: [Definite/Possible/Unlikely] Backup Plan: [If this fails/takes longer] Success Metric: [How we’ll know it’s done]