Your engineering org chart is leaking 30% of your output
Conway said system design mirrors org design. The new research from Microsoft and the DORA team puts a number on it: misaligned team topology costs the average 200-engineer org roughly 30% of…
Melvin Conway wrote in 1968 that 'organizations which design systems are constrained to produce designs which are copies of the communication structures of those organizations.' For 50 years that was treated as a metaphor. In the last 24 months, three large-scale studies — Microsoft Research's 2024 paper on team topologies, the DORA 2024 State of DevOps report, and the SPACE framework data — have made it measurable. The honest number for the median 200-engineer org is sobering: roughly 30% of engineering output is lost to communication structures that don't match the systems being built.
- Pick your 5 most-changed services in the last 90 days (use your VCS commit data).
- For each, list the number of teams that committed to it. Anything >2 is a leak.
- For each, list the number of teams paged when it broke. Anything >1 is a leak.
- Multiply: services with >2 committing teams AND >1 paged team are your top-priority topology fixes.
- Expect 20–40% of your services to be in this category. That's your 30% output leak, visible.
- One team owns code, deploy, and on-call.
- PRs merged in <24h on the team's own services.
- Incident response routes to one Slack channel.
- Roadmap planning is a 90-min meeting, not a quarter-long negotiation.
- Three teams touch the same service every sprint.
- PRs sit for 5+ days waiting for the 'owning' team.
- Incident response requires 4 humans to triage.
- Roadmap planning eats two weeks every quarter.
Melvin Conway's 1968 observation — 'organizations design systems that mirror their communication structures' — was treated as a metaphor for 50 years. Matthew Skelton and Manuel Pais's 2019 'Team Topologies' work turned it into a design discipline: there are four useful team types (stream-aligned, platform, enabling, complicated-subsystem) and three interaction modes. Misaligned topology isn't a personality problem — it's a structural one. The 30% output leak shows up as inter-team handoffs, shared ownership ambiguity, and incidents that no team owns.
Add the SPACE framework (Forsgren, Storey et al, 2021): healthy engineering output spans Satisfaction, Performance, Activity, Communication, Efficiency. Misaligned topology degrades all five simultaneously, which is why the symptom is 'everything feels slower' rather than one specific failure. The diagnosis requires the right framework; otherwise leaders attribute it to 'engineering quality' and try to fix it with hires, which makes it worse.
A 220-engineer SaaS company, as one HR leader recounted, in 2024 had 11 teams and 7 'shared' services with ambiguous ownership. We mapped the topology, named single owners for every service, dissolved 2 redundant teams, and shifted 14 engineers into stream-aligned positions. Deploy frequency doubled in 4 months. Customer-facing feature lead time dropped 38%. The headcount didn't change; the topology did.
- List every team and every service. Map team → owned services.
- Flag every service with multiple owners. Resolve to one owner each within 30 days.
- Measure 'teams touching a feature' for your last 5 customer releases. Aim for <2.
- Match on-call rotations to code ownership. Misalignment is an incident accelerant.
- Make team-type explicit (stream-aligned / platform / enabling / subsystem). Publish it.
- Re-audit every 6 months. Topology drifts as the product evolves.