Most teams assume you have to slow down to become more stable. Nathen Harvey, who leads the DORA team at Google Cloud, says a decade of research says the opposite is consistently true.
He walks through why throughput and stability move together rather than trading off, what the four DORA metrics actually measure (and don't), why "your CEO doesn't care about your deployment frequency," and his blunt advice for adopting the framework: don't trust DORA — prove it in your own organisation first.
Chapters
- 0:01:01 — Nathen's path from Chef Software to leading DORA
- 0:04:12 — The surprising finding: speed and stability move together
Either your software delivery is fast and stable, or it's slow and unstable. Our data just shows that slowing down doesn't actually make you more stable.
- 0:05:35 — Why smaller batches improve delivery and wellbeing at once
- 0:08:02 — The four keys, and what they're really measuring
- 0:10:47 — The most underused DORA resource: the core model
- 0:12:59 — Why "your CEO doesn't care about deployment frequency"
Your CEO doesn't care about your deployment frequency. You shouldn't trust DORA — you should prove it.
- 0:15:48 — Moloco's story: going beyond the four keys
- 0:18:20 — Why real-time isn't always the right expectation
- 0:26:11 — Starting small: how to actually bring DORA into an org
If you can't change the way one team works, you can't have any expectation that you're going to change how the entire organization works.
- 0:29:54 — AI as DORA's next research frontier
Key takeaways
- Throughput and stability aren't a trade-off: DORA's data consistently shows elite teams are both fast and stable — slow, careful delivery doesn't reliably produce more stable systems.
- Shipping smaller, incremental changes — often via feature flags — improves delivery performance and team wellbeing at the same time, including less burnout and higher job satisfaction.
- The four DORA metrics are trailing indicators of engineering capability but leading indicators of business outcomes — they're the pathway to results a CEO cares about, not the destination.
- Harvey's advice on trusting the framework is blunt: "you shouldn't trust DORA, you should prove it" — validate the connection between delivery metrics and your own business outcomes in your own context.
- Moloco won a DORA Award for going beyond the four keys by investing in FinOps, saving millions annually in cloud costs while also cutting ML training time — again showing cost and speed aren't necessarily opposed either.
- The single biggest leadership mindset shift DORA points to is transformational leadership: set the goals, then trust the team closest to the work to figure out how, rather than dictating the how from above.
