The short version
Five minutes, 30 questions: how much people use AI and would like to, 18 concrete obstacles in their own work, 8 beliefs about AI, and the one thing they'd fix. Every statement is answered on the same row, from "Don't agree" to "Agree", and agreeing with a barrier counts as held back.
Each barrier belongs to one of six areas and names who can remove it, so the report points to the people who can act: it says whether the problem is mostly skills, conditions or motivation, lists the top barriers with their owners, and shows every department next to everyone.
Six areas, four owners
"People aren't using AI" has different fixes depending on why. A missing tool is IT's to fix; an unclear rule, Leadership's; no time, the manager's; no confidence, Training's. So each barrier carries an area and an owner.
Before the barriers, two questions set the scene: "How often do you use AI in your work?" (Never, Tried it once or twice, A few times a month, A few times a week, Every day) and "How much would you like to use AI in your work, compared with now?" (Much less, A bit less, About the same, A bit more, Much more). The report groups people by the first into "Rarely or never", "Monthly" and "Weekly or more".
How the statements work
- One row for everything. Every barrier and belief is a statement with "How much do you agree?" under it and the same 1–5 row: Don't agree, Tend to disagree, Not sure, Tend to agree, Agree. Only the two ends are named on screen; the middle labels show on hover and to screen readers.
- Held back means 4 or 5. For each statement the report shows the share held back and the average; for each area, the share of its answers at 4–5.
- The Barrier index is the average answer to the 18 barriers, with "Don't agree" as 0, "Not sure" as 50 and "Agree" as 100. Lower is better, and it has no bands.
- The last two questions are "If we could fix one thing to help you use AI more, what would it be?", one choice from 9 (Better AI tools or a paid plan; A safe, approved way to use AI with our data; Clear rules on what's allowed; Time set aside to learn and try AI; Backing from my manager; Training for my role; Examples and templates for my work; Someone to ask for help; Nothing, I have what I need), and "Which department are you in?", optional and last, so it can't make people careful with their earlier answers.
Eight beliefs
Barriers are about the work; beliefs are about AI. Seven of them hold people back one way or the other, and the "agree" and "disagree" statements alternate on purpose, so nobody gets through the block by agreeing with everything. One is shown for context and never ranked: it can push people either way.
One belief gets a check: the share who think most colleagues use AI regularly is set against the share who say they use it weekly or more. When the gap is 10 points or more, the report says so: people often underestimate their colleagues.
Where to invest first
The areas roll up into three parts, after the model behind the survey: skills (Know-how), conditions (Tools, Rules & data, Support and the two time statements) and motivation (Usefulness, the habit statement and the beliefs). The part with the highest share held back is the verdict:
Under the verdict, the report lists the top 5 barriers, each with who can fix it, and suggested actions for the three areas that hold people back most:
What the report shows
- Top summary: the Barrier index, the biggest barrier area with its owners, the most wanted fix with its owner, and AI use (the share using AI weekly or more, and the share who want to use it more).
- Where to invest first, the top 5 barriers and barriers by area, each area opening on its statements.
- Who's stuck where: a heat map of the six areas and the beliefs, by department or by how much people use AI, next to everyone. Cells show the share held back, from "Under 15%" to "60% or more", and every group shows from its first answer.
- What people believe and what would help most, the one fix each person asked for, with its owner.
- Trends: the Barrier index, the verdict and each area's share held back, round by round, and this round's top barriers over time. Lower is better.
The department list is set per round on the report's Configuration tab; leave it empty and the question isn't asked.
The model behind it
The areas follow COM-B (Michie, van Stralen and West, 2011), a widely used model of what a change in behaviour needs: capability (Know-how), opportunity (Tools, Rules & data, time, Support) and motivation (Usefulness, habits, beliefs). The statements are our own, written for this survey, not a validated scale like the AI Anxiety Scale. After the first real runs we can check that each area's statements move together before claiming more.
It complements the AI Anxiety Scale: that one measures how people feel about AI in general; this one asks about concrete obstacles in their work.
Who sees what
- It can run anonymously, and the one-click launch does: a shared password with no names. Advanced setup can ask for a name or an email, or use Google sign-in, and then the Responses tab names people, department included.
- Every group shows from its first answer. There is no minimum size for a department column, so a department of one is that person's answers. The public link, if one is made, shows the Insights with the heat map, as aggregates.
Limits
- Not a validated scale. The six areas follow a well-known model, but the statements are new. The "Research-based" badge means the framing, not a tested instrument.
- Rankings always fill. The top 5 barriers, the top 3 areas and the top 3 beliefs are relative: they show even when every share is low. Read them with the shares.
- Shares mix units. Statements and areas count answers; AI use, wanting more and fixes count people.