Short answer
An AI adoption survey asks staff how often they use AI tools for work, for which tasks, how much time it saves, how confident and careful they are with it, and what holds them back. This template has 18 questions with fixed response scales, takes about seven minutes, and includes scoring rules so you can track change after training or a rollout.
- Fixed scales let you compare a baseline with later surveys.
- Ask about personal AI accounts without blame, or people will not answer honestly.
- Report team results only where five or more people responded.
- Self-reported time savings are indicative, not measured.
Send this survey to everyone in scope, including people who do not use AI; their answers matter most. Tell people why you are asking, what will happen with the results and that answers are confidential. That is good practice and, because responses can identify people, it reflects the Privacy Act’s rule that people should know the purpose when you collect their information (Privacy Act 2020, IPP 3).
The AI adoption survey
A. How you use AI
- In the last month, how often did you use an AI tool for work? Never / Less than weekly / Weekly / Several times a week / Daily
- Which AI tools did you use for work? Tick all: [approved tool 1] / [approved tool 2] / AI features in our other software / a personal AI account / other / none
- In the last month, did you use a personal, non-work AI account for work tasks? No / Once or twice / Regularly / Prefer not to say
- What do you use AI for? Tick all: drafting emails / summarising documents / meeting notes / research / working with spreadsheets or data / customer replies / writing or checking code / images / other
B. What it is worth
- In a typical week, how much time does AI save you? None / Under 30 minutes / 30 to 60 minutes / 1 to 3 hours / More than 3 hours
- “AI has improved the quality of my work.” 1 Strongly disagree / 2 Disagree / 3 Neutral / 4 Agree / 5 Strongly agree
- What one task would you most like AI to help with that it does not today? Open text
C. Confidence and safe use
Use the same 1 to 5 agreement scale for each statement.
- “I know which AI tools I am allowed to use for work.”
- “I know what information I must not put into AI tools.”
- “I check AI output against sources before I use it.”
- “I can get useful results from AI for my regular tasks.”
- “I know who to tell if AI produces something wrong or harmful.”
- “I am comfortable with how AI is being introduced in our team.”
D. What holds you back
- What stops you using AI more? Tick up to three: not sure what is allowed / do not trust the output / checking takes too long / not relevant to my work / no time to learn / the tool I need is not available / worried about the effect on my job / other
- What support would help most? Tick up to two: short how-to sessions / examples for my role / a shared prompt library / one-to-one help / clearer rules / none
E. About you, for grouping only
- Team: [list teams; teams under five people are combined]
- How long have you used AI tools for work? Not yet / Under 3 months / 3 to 12 months / Over a year
- Anything else you want to tell us? Open text
How to score the results
| Measure | How to calculate | Use it to |
|---|---|---|
| Adoption rate | Share answering weekly or more often to Q1 | Track uptake over time |
| Personal account use | Share answering “once or twice” or “regularly” to Q3 | Size the privacy risk and the need for approved tools |
| Safe-use index | Average of Q8, Q9, Q10 and Q12 (1 to 5) | Decide whether rules and training are landing |
| Confidence index | Average of Q11 and Q13 | Spot teams that need hands-on support |
| Estimated time saved | Q5 band midpoints: 0, 0.25, 0.75, 2 and 4 hours a week | Indicative only; confirm with timed tasks |
| Top barriers | Count of ticks in Q14 | Choose the next fix |
Report team-level results only where five or more people answered. Below that, individuals can be identified.
How to run it
- Set it up in your usual form tool. Microsoft Forms, Google Forms or similar; turn off name collection.
- Explain the purpose in the invitation. Planning training and support, not assessing individuals.
- Run the baseline before any change. Then repeat about three months after training or a rollout, as described in measuring adoption after training.
- Add usage data where you have it. Some business plans include it; ChatGPT Business, for example, lists usage analytics (OpenAI).
- Share results and actions with staff. People answer the second survey if they saw the first one change something.
Illustrative results
Illustrative example: a 40-person Wellington professional services firm surveyed staff in July 2026, ran training in August using its training needs assessment, and surveyed again in October. All figures are invented for illustration.
| Measure | July baseline | October follow-up |
|---|---|---|
| Adoption rate | 38% | 71% |
| Personal account use | 30% | 8% |
| Safe-use index | 2.9 | 4.2 |
| Confidence index | 3.1 | 3.8 |
| Most-ticked barrier | Not sure what is allowed | Checking takes too long |
The barrier shift was the useful finding. Once rules were clear, the cost of checking output became the main complaint, so the next step was a shared prompt library with known-good examples rather than more general training.
Common mistakes
- Only surveying users. Non-users explain the barriers.
- Naming and shaming. Questions about personal accounts are only answered honestly when there is no blame. Employers and employees must deal with each other in good faith (Employment NZ); a survey used against people breaks that trust quickly.
- Changing the questions each round. Then nothing is comparable.
- Treating time saved as fact. People overestimate and underestimate. Measure real tasks before putting a number in a business case.
Next step
Turn the barriers into a schedule with the AI learning plan for teams, and if personal account use is high, review managing AI subscriptions and access. For a structured rollout, see rolling out AI to a small team or our AI training for business teams.
Sources
- Privacy Act 2020, New Zealand Legislation, version as at 1 October 2026
- Good faith, Employment New Zealand, last modified 25 September 2026
- ChatGPT pricing, OpenAI, accessed 19 October 2026
Written by the Levy Street team in Wellington. Last updated 19 October 2026.