How do you build high performance teams?

• software delivery performance (DORA, puppet labs)
• You don’t improve deployment speed by focusing on deployment systems.
• Getting software in front of customers quickly is the real goal.
• Deliver software faster and safer.
• If we help teams improve they will consume more of the product.
• How do you provide conditions for those metrics to improve?
• Who’s paying for this?
• What business metrics are we really interested in?
• Metrics is all about industrialization, but you’re trying to make groups of humans to work?
• Je ne sais quoi
• 25% of new code going to production at google is written by machines (scripted tool applying changes to code?) - is this productivity?
• are you using enough gen ai and how is it boosting productivity? Should we see 10x performance? A lot of orgs prioritize using ai, but to what end? FOMO? We need better reasons.
• what is the impact of AI? Feeling more productive at individual level, time spent on toilsome work is unchanged, amount of time doing valuable work goes down - are you delivering same value faster? Use the extra time for something new (not valuable work), at team level ai improves review process, doc quality goes up (improves doc utility through ai)
• When more ai is used software delivery falters (oh shit moments increase)
• what does it mean to be productive? Compare ourselves to our competitors?
• Software engineering is creative work and when we treat it as industrial process we fail (feature factory has much lower performance vs. teams who understand why, users, etc)
• continuous improvement, fund what works, what lessons can be learned (but remember they work under different contexts)
• story points?!
• how many experiments did you launch?
• big leverage points are people, not process? Team composition, advocate for different positions, throw away their beliefs
• Morale, measure? Dipping? Don’t ask but know!
• can metrics be harmful? Metrics educate a non technical team?
• institutionalizing change, what do we need to change? The change is what we need
• Measuring and instrumenting is very expensive, Amazon gave up, ROI? Survey is anonymous, broad based trends, can’t tell who’s participating but draws a bunch of connections - team ran a survey, presented results, that’s your opinion, two years to build a dashboard, nothing changed it didn’t matter
• an entire org capturing metrics built over decades, just take our findings and replicate them in your context, use them as hypothesis as a change you will make, you don’t need precise data
• metrics aren’t static
• You cannot succeed without talking to developers
• Do something small, talk to your team

No file, test, or entry point is named; begin by determining whether this open-ended set of notes is intended to become a repository article or another deliverable. The issue does not define what done looks like or provide concrete acceptance criteria.