The 174 AI literacy assessment rubric

The assessment most companies offer ends with a sales call. Ours ends with a report you can forward to your CHRO, your COO, or your CEO. To make that report defensible, the rubric behind it has to be public. This is that rubric — every question, every option weight, every bucket threshold, and the logic that turns answers into a recommended rollout shape.

If you’d rather just see your own number, the 10-minute assessment is the fastest way. If you want to read the methodology first — or you’re the kind of buyer who reads methodology before clicking anything — this page is for you.

Why three dimensions

Most AI-literacy assessments measure one of two things: tool usage (“does your team use ChatGPT?”) or sentiment (“how excited are people about AI?”). Both are misleading. Tool usage misses whether the work is any good; sentiment misses whether anything is shipping.

The three dimensions in the 174 rubric — adoption, capability, and governance — exist because they’re the smallest set that captures what actually matters for a mid-market rollout:

• Adoption is whether AI is in the work, day-to-day, across the people who would benefit. It’s the breadth signal.
• Capability is whether the work being done with AI is actually good — prompts that get reused, output that gets evaluated, agents that ship. It’s the depth signal.
• Governance is whether the program can survive scale — policy, quality controls, executive sponsorship. It’s the durability signal.

Drop any one and you can construct an organization that looks healthy on the other two but falls over the moment a rollout actually begins. We considered splitting governance into “policy” and “controls” — they’re meaningfully different — but in practice mid-market buyers think of them together, and conflating them costs us very little. Three dimensions, scored 0–100 each, bucketed into Emerging / Developing / Mature.

The questions

All 12 questions are below. Each option carries a score weight; the total possible score per dimension is the sum of each question’s max-option score. A buyer’s percentage in each dimension is (actual / max) * 100, rounded.

Adoption

A1. What share of your team uses AI tools in their day-to-day work?

Single select. The breadth-of-use signal.

Option	Score
Less than 10%	0
10–30%	25
30–60%	60
More than 60%	100

A2. Which AI tools are in active use across your team?

Multi select (up to 6). Each selection contributes — “None of the above” is its own option scored at 0 to keep the math honest. The mix matters: a team running custom agents alongside general assistants is signalling a different maturity than one running only ChatGPT.

Option	Score
ChatGPT / Claude / general assistants	20
Copilot for code (GitHub, Cursor, Windsurf)	20
AI features inside existing SaaS (Notion AI, Slack AI)	15
Custom prompts or GPTs in shared workspaces	20
Internal agents / automations (n8n, Zapier AI, custom)	25
None of the above	0

A3. How often does your team use AI for real work (not experimentation)?

Single select. Frequency-of-real-use, distinguished from playing-with-it.

Option	Score
Rarely or never	0
A few times a month	25
Weekly	60
Daily	100

A4. How would you rate your team’s satisfaction with current AI tools?

Single select. Captures whether adoption is durable or about to churn.

Option	Score
Frustrated — outputs aren’t reliable	0
Mixed — some wins, some misses	33
Generally positive	66
Strong — it’s changed how we work	100

Capability

C1. How would you rate your team’s prompting skill?

Single select. Honest answer beats aspirational.

Option	Score
Most people just type a question	0
Some patterns, but ad-hoc	33
Documented prompts that get reused	66
Evaluated, versioned, reviewed prompts	100

C2. How does your team evaluate AI output before relying on it?

Single select. Eyeballing and structured evaluation are two different rooms.

Option	Score
We don’t — we eyeball it	0
Spot checks by a senior person	33
Informal rubrics or checklists	66
Structured evaluation with documented rubrics	100

C3. How often does your team use AI in multi-step / agentic workflows?

Single select. Beyond single prompts — chained, automated, or agent-like.

Option	Score
Never	0
Experimenting	33
A few production workflows	66
It’s a core part of how we work	100

C4. How does your team currently learn new AI techniques?

Multi select (up to 5). Captures whether learning is structured or entirely incidental.

Option	Score
YouTube and Twitter / X threads	10
Internal Slack channels and word of mouth	15
Courses or certifications	25
Internal lunch-and-learns or workshops	25
A formal AI literacy program	25

Governance

G1. Do you have a written AI usage policy?

Single select. The single highest-leverage governance instrument.

Option	Score
No	0
Drafting one	33
Yes — published but not enforced	66
Yes — published, trained, enforced	100

G2. Who owns AI enablement in your organization?

Single select. Ownership is the difference between “we should do AI” and “we are doing AI.”

Option	Score
No one in particular	0
IT or Security — mostly risk management	33
L&D or People Ops	66
A dedicated AI / transformation lead	100

G3. Are there quality controls for AI output that ships externally?

Single select. Customer-facing copy, code, contracts, decisions.

Option	Score
No controls	0
Manager review only	33
Documented review steps	66
Documented + enforced + audited	100

G4. What’s the level of executive sponsorship for AI literacy?

Single select. Without an exec sponsor, the program lives until the next budget review.

Option	Score
No exec sponsor	0
Verbal support, no budget	33
Budget allocated, no clear program	66
Funded program with exec accountability	100

How scoring works

Each dimension is computed independently:

1. Sum the scores for the answers given to the questions in that dimension.
2. Sum the maximum possible scores for the same questions.
3. Divide and multiply by 100. Round to the nearest integer.

For Adoption (questions A1 + A2 + A3 + A4), the maximum is 100 + 100 + 100 + 100 = 400. A response earning 0 + 40 + 25 + 33 = 98 produces an Adoption score of round(98 / 400 * 100) = 25.

For Capability and Governance, the same arithmetic applies with their respective maxima.

The overall score is the rounded average of the three dimension scores. There is no weighting — we treat the dimensions as equally important on purpose, because the failure mode of weighting is letting a single strong dimension hide weakness elsewhere.

How buckets work

Score	Bucket
0–39	Emerging
40–69	Developing
70–100	Mature

The thresholds aren’t tuned to a normal distribution — they’re tuned to the rollout decisions a buyer needs to make. A score below 40 in any dimension means the program for that dimension is not yet running. A score between 40 and 69 means the program is running but inconsistent. A score of 70 or higher means the program is durable and ready to scale or deepen.

How recommendations are generated

For every (dimension, bucket) pair, the rubric carries two pieces of copy:

• The highest-leverage gap. What’s actually limiting progress in that dimension at that maturity. Phrased as a diagnosis, not a complaint.
• The recommended next move. What 174 would suggest you do first. Phrased as a single concrete action, not a workshop list.

These pairs aren’t generated dynamically — they’re authored. We update the copy when we learn something new from a real engagement; we don’t update them per-buyer. The full mapping is in src/data/assessment.ts in the marketing repo if you want to read every variant.

The overall rollout shape is computed from your overall score plus your weakest dimension:

• Below 40 overall: a 1–10 seat pilot, in the team most ready to move.
• 40–69 overall: a department-wide rollout, starting with the weakest dimension.
• 70+ overall: org-wide concierge — the program scales now.

How to interpret your own score

Three honest principles:

1. The rubric rewards honest answers. Picking aspirational options (“we’re drafting one” when you haven’t started) inflates the bucket and recommends the wrong rollout shape. The point of the assessment is to see clearly, not to feel good.

2. Low scores in Emerging are useful, not damning. A 25 on Adoption is a clear instruction: visibility and access first. A 25 on Capability is a clear instruction: curriculum first. The bucket is your starting line, not a grade.

3. Mature in one dimension, Emerging in another is the most common shape. Mid-market companies routinely have strong adoption and weak governance, or vice versa. The assessment surfaces that asymmetry on purpose — most rollout failures are governance failures dressed up as adoption successes.

If you’ve taken the assessment and want to retake it after a few months of work, the URL is the same. The report you generate is dated, so you can put two reports side by side and see the actual lift.