Improve developer experience by making important work faster and easier to complete without sacrificing quality or wellbeing. Measure those outcomes together, then use small, evidence-based changes to remove friction. Faster coding or more activity alone does not show that engineering performance has improved.
What does developer experience have to do with engineering performance?
Developer experience is the set of conditions that shapes how people move through engineering work: how quickly they can make progress, how much friction they encounter, the quality of what they deliver, and whether the work is sustainable. It is therefore a system-level concern, not simply a question of whether developers like a tool.
Microsoft Research’s EngThrive model organizes productivity around Speed, Ease, and Quality, with Thriving as a wellbeing guardrail. Its authors—Brian Houck, Tim Bozarth, David Liu, and Dean Carignan—describe the aim this way: “EngThrive organizes productivity around three dimensions – Speed, Ease, and Quality – with Thriving as a guardrail to ensure developer wellbeing improves alongside performance.” The model pairs outcome-oriented North Star measures with diagnostic measures, drawing on system telemetry and developer surveys for context. Microsoft Research, EngThrive.
EngThrive was developed and deployed at Microsoft; it is a useful measurement model, not proof that every organization should copy its exact measures. The practical lesson is to assess performance across connected dimensions rather than optimizing one number in isolation.
Recommended Free Tools
#1 Best Overall
What should you measure?
Choose a small set of outcome measures tied to work that matters, then use diagnostic signals to understand why those outcomes change. The categories below are suggested starting points, not a universal metric set or a benchmark. Define them around your own workflows and validate that they reflect meaningful work.
| Dimension | Possible signals | What to check |
|---|---|---|
| Speed | Time or flow through a meaningful developer workflow | Look at how work moves at team or system level, rather than treating individual activity as the outcome. |
| Ease | Workflow completion success, avoidable waits, repeat support requests, and reported friction | Find where developers get blocked, repeat steps, or need help to finish routine work. |
| Quality | Reliability and change outcomes, including stability signals relevant to your organization | Check whether faster progress is accompanied by acceptable delivery and operational outcomes. |
| Thriving | Developer wellbeing and satisfaction | Treat wellbeing as a guardrail, not a benefit to trade away for a local speed gain. |
| Context | Diagnostic telemetry combined with developer survey feedback | Use these signals to interpret outcomes; neither instrumentation nor opinion alone tells the whole story. |
Do not use lines changed, commits, tasks closed, or tool adoption as stand-ins for productivity unless you can show that a particular measure reflects valuable outcomes in your context. Activity counts can rise while work becomes harder, quality suffers, or effort shifts to testing, review, and operations.
How do you turn measurement into improvement?
Start with a workflow developers repeatedly struggle to complete. Establish its baseline, state a specific hypothesis about the friction you intend to remove, make a focused change, and check what happened. DORA recommends this iterative approach: “Taking an experimental approach to continuous improvement remains essential for modern teams.” DORA Research: 2024.
- Choose a recurring workflow. Select a task that matters to delivery and causes repeat delays, confusion, support requests, or unnecessary handoffs.
- Set a baseline. Record the relevant outcome and diagnostic signals, and ask developers about their experience completing the workflow. Be clear about the workflow boundary so comparisons remain meaningful.
- Write a testable hypothesis. For example: “Making the self-service setup instructions clearer will increase successful first attempts and reduce avoidable help requests.” This is a hypothesis to test, not a guaranteed result.
- Make one focused change. Improve the instructions, make feedback more actionable, or remove a recurring dependency on an enabling team. Keep the change narrow enough to connect it to the workflow being assessed.
- Re-measure and decide. Compare the outcome and developer feedback with the baseline. Keep, revise, or reverse the change based on whether it helped, caused harm, or moved friction elsewhere.
Small loops make it easier to learn before a local fix becomes a broad standard. When an apparent gain in one team coincides with more work for another, include the affected workflow in the next measurement rather than counting the first gain as a net improvement.
How should you assess an internal developer platform?
An internal developer platform can make common work easier to complete through self-service workflows, clearer task outcomes, and fewer recurring dependencies on an enabling team. But platform adoption is not itself proof of better engineering performance. DORA’s 2024 findings say platforms can improve individual, team, and organizational performance while also potentially decreasing throughput and change stability. The results can differ by outcome, so monitor tradeoffs rather than assuming every measure will improve. DORA Research: 2024.
Evaluate platform work against the workflows and outcomes that matter to your organization:
- Developer independence: can teams complete common tasks without avoidable handoffs?
- Workflow completion: can developers complete the task successfully, and where do they still get stuck?
- Feedback quality: does the platform make results and next steps understandable?
- Delivery outcomes: what happens to speed, throughput, change stability, and other relevant reliability signals?
- Different team needs: do teams with different workflows and constraints benefit, or does the platform fit only a subset?
Begin with self-service workflows that remove recurring dependencies and make outcomes legible. Treat the measures as a way to discover where the platform helps and where it creates new friction—not as a score that rewards adoption on its own.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How should you evaluate AI tools?
Evaluate AI in the full delivery system, not just at the point where code is produced. Consider effects on coding, testing, review, security, deployment, and the effort required to understand and maintain generated work. A faster first draft may not improve overall performance if it increases downstream review or reliability costs.
Free tools Windows power users keep installed
One-click scans. No signup required.
Best Value
DORA’s 2025 report characterizes AI as an amplifier of organizational strengths and dysfunctions. That framing makes existing engineering conditions part of the evaluation: individual gains cannot, on their own, establish that organizational performance has improved. The report draws on more than 100 hours of qualitative data and survey responses from nearly 5,000 technology professionals worldwide; those findings inform hypotheses, not guaranteed results for every organization. DORA 2025 State of AI-assisted Software Development Report.
How can developer feedback become useful evidence?
Ask developers about specific workflows and recurring points of friction, then interpret those answers alongside system measures. Feedback can reveal confusion, hidden handoffs, or avoidable effort that telemetry does not explain; telemetry can help show whether a reported problem affects workflow outcomes. Neither should be treated as a substitute for the other.
Google Research describes a quarterly, large-scale developer survey at Google that had been running since 2018, with lessons and refinements accumulated over six years. That example supports treating survey practice as something to maintain and improve over time, rather than a one-off satisfaction poll. It does not establish that another organization needs the same survey design or cadence. Measuring Developer Experience with a Longitudinal Survey.
What can published research tell you—and what can’t it?
DORA’s 2024 research covered more than 39,000 professionals across organization sizes and industries globally. Its results are useful evidence for forming and testing improvement ideas, not a promise that a particular intervention will produce the same outcome in every organization. The 2024 and 2025 figures describe different research efforts and should not be combined as if they were a single sample. DORA Accelerate State of DevOps 2024 Report.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
There is no universal threshold in these sources for “high engineering performance.” Set local definitions for meaningful outcomes, explain how you measure them, and look for changes across speed, ease, quality, and wellbeing. A change is not a complete improvement if it merely shifts cost to another stage of delivery or makes developers’ work less sustainable.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
