Developer experience survey: 14 questions on where engineers lose time
A DevEx survey your engineers answer about their own last four weeks: how long they wait, how hard the system is to understand, and whether they get time to think. Answer it below, or open it in the editor and put your team names in.
- 14questions
- about 4 minto answer
- 5-point agreeplus two open answers
- 4 weeksof one person's work
What a developer experience survey measures
Where an engineer's week leaks away, asked of the engineer, about the last four weeks.
Three blocks, one outcome. The blocks follow the three dimensions of the DevEx framework: how fast feedback comes back, how much effort it takes to understand the system, and whether there is time to focus.[1] The wording is ours and every item asks about something that happened, not a general feeling. The overall rating comes last, so you can see which block moves it.
Your developers, not your users. People using the product you build answer a user experience survey instead. The fourteen questions below are already built, so you can create a free survey from them with your own team names.
The 14 DevEx survey questions
Every question, its answer scale, and what a low score points at. Copy one block or the whole set.
Your work
Two fields first, so every later answer can be cut by team and read against how much coding the person did.
- Which team do you mostly work in?Swap this for a pick list of your real team names. Free text lets one person's spelling identify them.
- In the last four weeks, how much of your time went on writing, reviewing or shipping code?Exposure, not opinion. A lead who coded for two days reads the build questions differently from someone who coded every day.
Waiting on feedback
How long a developer waits to learn whether their change works: the machine, the reviewer, the pipeline, the release.
- A local build and test run is quick enough that I do not switch to something else while I wait.Disagreement means context switching while builds run, which costs more than the minutes on the clock.
- Review of my changes usually starts within one working day.A review queue problem. Check it against your own review timestamps before blaming reviewers.
- When the pipeline fails, the output tells me why without digging.Low scores usually mean flaky tests or logs nobody can read, not slow infrastructure.
- Getting a merged change into production needs no manual chasing.Manual release steps, approvals or a deploy only one person knows how to run.
Understanding the system
The effort it takes to know enough to change something safely.
- I can work out how a service I do not own behaves without having to ask someone.Low here and high on question 8 means the docs exist but do not cover other teams' services.
- The documentation I rely on is accurate enough to act on.Stale docs are worse than none: people act on them. Ask which ones in the open answer.
- Getting access to a repository, tool or environment I need does not stall my work.Often the cheapest fix in the set: a request process nobody owns.
Time to focus
Whether the week leaves room for the kind of work that needs a long stretch of attention.
- Most days I get at least one two-hour stretch without meetings or interruptions.Two hours is the test because most real engineering tasks do not fit in the gaps between meetings.
- Unplanned work rarely breaks up the work I had planned for the day.Low scores point at on-call load, support rotas or urgent requests landing on the same people.
- In the last four weeks, which of these cost you the most time?Count the ticks per option. This is the ranked fix list; "Nothing significant" sits first so a good month costs one tap.
In your words, then the verdict
One open answer, then the rating everything above explains.
- If you could remove one thing that slows you down, what would it be?One thing, not a list. Forty single answers can be grouped and counted.
- Overall, how would you rate your developer experience over the last four weeks?The outcome. Read the three blocks above against it to see which one moves it.
Nine items share one agree scale: Strongly agree, Agree, Neither agree nor disagree, Disagree, Strongly disagree. Keeping one Likert scale is what lets the blocks be averaged and compared across quarters.
Scoring developer experience
One headline, three block averages, and your own last quarter as the only comparison.
The headline. The share of answers to question 14 that said good or very good. Track it quarter to quarter.
The blocks. Score the agree items 5 down to 1 and average each block: questions 3 to 6, 7 to 9 and 10 to 11. The lowest block tells you where to start, and the ticks on question 12 rank the fixes inside it.
No benchmark. Nothing comparable is published for these items, so compare a team with itself. Put the survey beside your system data, such as build times and review wait, because the paper behind the framework recommends reading both together.[1]
| If this block is lowest | What it usually means | Where to look first |
|---|---|---|
| Waiting on feedback (Q3 to Q6) | Slow or flaky builds, a review queue, manual release steps | Pipeline duration and failure logs, review start times |
| Understanding the system (Q7 to Q9) | Knowledge held by a few people, stale docs, access requests nobody owns | The services people named in question 13 |
| Time to focus (Q10 to Q11) | Meeting load and unplanned work landing on the same people | The on-call and support rota, recurring meetings |
Sending the developer survey
Who, when, and how anonymous.
Everyone who writes, reviews or ships code, including platform, QA and site reliability engineers. Engineering managers who rarely code can answer, and question 2 lets you read them separately. Staff outside engineering rating the tools they are given answer a technology survey instead.
Quarterly, in the same week each quarter, with a one-week window. Avoid a week with a major incident or a release freeze. Tell people what changed after the last round; that is what keeps them answering.
Anonymous, reported by team, and only for teams with five or more answers, with no role or seniority field, because a few details together can identify one person.[4] For more than your engineers, see the other technology survey templates.
Then hand off what this does not cover. One tool up for renewal belongs in a software evaluation survey. Complaints about laptops and tickets are IT support survey questions for everyone.
And the sprint is its own survey. How the last two weeks went for one team is a retrospective survey, sent before the retro.
Developer experience survey FAQ
What is a developer experience survey?
A short, regular survey engineers answer about the friction in their own work: how long they wait for builds and reviews, how hard it is to understand the systems they change, and whether they get uninterrupted time. It is the self-reported half; your tooling data is the other.
What questions should a DevEx survey ask?
Questions about concrete, recent events rather than general feelings: did the build finish before you switched tasks, did review start within a day, could you find how a service works. Group them by where the time goes, keep one agree scale, add one open answer, and ask the overall rating last.
How often should you run it?
Quarterly suits most teams. That is often enough to see whether a fix worked. Keep the wording identical between runs, or the trend stops meaning anything.
Should a developer experience survey be anonymous?
Yes. Report results by team only, and only where five or more people answered. Use a fixed list of team names rather than free text.
How do DevEx, SPACE and DORA fit together?
DORA metrics measure delivery from system data. SPACE is a broader framework saying productivity has several dimensions, including satisfaction. The DevEx framework narrows that to three things developers feel directly: feedback loops, cognitive load and flow. This survey measures those three from the developer's side.
Michael Hodge, survey methodology and questionnaire design · Updated 22 September 2026 · How templates are reviewed
Sources (4)
- Noda, A., Storey, M.-A., Forsgren, N. and Greiler, M. (2023). "DevEx: What Actually Drives Productivity." ACM Queue 21(2). Named and described only; no item reproduced. doi.org/10.1145/3595878
- Forsgren, N., Storey, M.-A., Maddila, C., Zimmermann, T., Houck, B. and Butler, J. (2021). "The SPACE of Developer Productivity." ACM Queue 19(1). doi.org/10.1145/3454122
- DORA (DevOps Research and Assessment). "DORA research program." dora.dev
- American Association for Public Opinion Research. "Best Practices for Survey Research." aapor.org
Fourteen questions written for this page. The three driver blocks follow the dimensions named in the DevEx framework[1], which sits inside the wider SPACE view that productivity has several dimensions, satisfaction among them.[2] Delivery metrics of the kind DORA publishes are system measures and are not asked here.[3] No item comes from a licensed or proprietary instrument. Both papers are ACM copyright and are cited for their ideas only; the wording of every question is original.
Send it at the start of next quarter
Fourteen questions, about four minutes, and a baseline to measure every platform fix against. Put your team names in and send it.
Use this template