Introduction to Part 9: People, Teams, and Improvement Metrics
Every service depends on people who are motivated, supported, and free to improve their own work, so the measures in this part describe conditions and habits, and none of them is a way to rank or watch anyone.
The first eight parts of this book describe services, flows, and records. This part turns to the people who build and run them. Behind every metric stands a team that has to live with it, and the quality of the work cannot rise above the conditions the team works in. The Agile Manifesto's principles say it plainly: build projects around motivated individuals, give them the environment and support they need, trust them to get the job done, and let them keep a constant pace indefinitely. They also say that the team should reflect regularly on how to become more effective. The measures in this part are how those statements become checkable.
IT service management (ITSM), as codified by ITIL, approaches the same ground through its workforce and talent management, knowledge management, organisational change, and continual improvement practices. The two traditions agree on the essentials: people need clear purpose, safety to speak up, time to learn, and a say in how the work is organised. Where they differ, the difference is mostly vocabulary.
This part carries the heaviest ethical load in the book. Measures that touch people can slide into surveillance, create fear, and produce false answers. A single wrong use, such as ranking individuals by an engagement score or tracking hours online, can ruin years of trust and make every later measure useless. For that reason each topic here states its privacy rules, aggregates at team level, sets minimum group sizes, and bans the use of the data in individual evaluation. Each metric still comes with the way it gets gamed and a guardrail, because even well-meant measures about people are distorted when they become targets (see topic 1.2, Goodhart's law and the psychology of metrics).
The four lenses shape this part. A sustainable pace and motivated individuals are the aim of the measures. AI assistants change workload and skills, so ask about benefit and burden directly. Automation should be used to collect and remind, not to monitor people. Analytics reports at team level, with response counts and a minimum group size.
Topics in this part
- 9.1 Team health and developer experience: feedback loops, cognitive load, flow, team-owned health checks, and obstacle removal.
- 9.2 Satisfaction and well-being metrics: sustainable pace, psychological safety, workload strain, and how to ask without prying or diagnosing.
- 9.3 Collaboration and handoff metrics: handoffs, waiting, dependencies, real-time collaboration, shared knowledge, and stakeholder response time.
- 9.4 Skills and learning metrics: protected learning time, cross-training, onboarding, communities of practice, and lessons shared across teams.
- 9.5 Continual improvement metrics: follow-through on actions, effect checks, repeat issues, and capacity for improvement.
- 9.6 Organisational change and adoption metrics: aim clarity, outcomes against comparison, change load, voluntary uptake, and the difference between a practice and a ritual.
How these topics interrelate
Topic 9.1 (see topic 9.1, Team health and developer experience) describes the daily conditions: how fast feedback arrives, how heavy the load on the mind is, and whether people can concentrate. It introduces the rule that runs through the whole part, which is that the team owns its own health check and reads the results first. Topic 9.2 (see topic 9.2, Satisfaction and well-being metrics) goes to the deeper question of whether the pace can be sustained and whether people feel safe to say so, and it gives the strictest privacy rules in the book.
Topics 9.3 and 9.4 look at how work and knowledge move among people. Handoffs and dependencies (see topic 9.3, Collaboration and handoff metrics) show where work waits. Skills and learning (see topic 9.4, Skills and learning metrics) show whether knowledge is spread widely enough that the waiting and the risk fall. Both depend on conditions from the first two topics: a team without focus time or without safety to ask for help will not share what it knows.
Topics 9.5 and 9.6 close the part with improvement and change. Continual improvement (see topic 9.5, Continual improvement metrics) asks whether ideas become changes and whether the changes are checked. Organisational change (see topic 9.6, Organisational change and adoption metrics) asks whether a new way of working has taken hold for its intended reason or has spread only as a ritual. Both are tests of the conditions set out earlier: improvement needs protected time and safety, and change that ignores the pace and the people fails.
Several guardrail pairs follow. A team health score should be read with the obstacle list and what was done about it. A well-being pulse should be read with team-level workload facts, never individual hours. A learning measure should be read with evidence of capability in use, so that attendance does not stand in for skill. An adoption figure should be read with outcomes against a comparison group, so that form is not mistaken for substance.
This part draws on the measures of earlier ones: flow and planning (see topic 2.5, Delivery forecasting and planning metrics), operational load (see topic 2.9, Operational work and unplanned work), toil and on-call (see topic 5.7, Toil and on-call metrics), and the experience of internal users (see topic 6.4, User and employee experience). It leads into the last two parts: AI changes what many of these measures mean, and a metrics programme is itself a change that needs to be introduced with the care this part describes.