The AI Readiness Assessment

Most AI readiness assessments ask what you've bought. Tool count, pilot count, how enthusiastic leadership is, how modern the stack looks. A firm can score well on every one of those questions and show no economic change at all, which is why those assessments produce certificates instead of decisions.

This one asks what changed. Twelve questions: one honest guess, four about your financials, and seven about how the firm actually runs. It places you on The Delivery Model Ladder — the four-stage AI adoption framework I use to locate service firms by economic signature — and hands back a plan built only from the moves you haven't made.

The AI Readiness Assessment is a twelve-question diagnostic that places a $1–20M service firm on the Delivery Model Ladder by which of four numbers moved — cost to deliver, retention, margin, and revenue per person — rather than by which tools it bought. It takes under three minutes, runs entirely in your browser, stores nothing, and every rule it applies is stated in full on this page. If you don't track the numbers it asks about, it says so, and that becomes the finding.

What is an AI readiness assessment, and what should it actually measure?

The genre has a standard shape: a questionnaire about culture, tooling, and intent that returns a maturity score and a sales call. The problem isn't the questionnaire. It's the inputs. Enthusiasm, pilots, and subscriptions are all measures of adoption, and adoption and economic change are different things. In the delivery data I've reviewed across a franchise network of 150+ units and growing, firms with heavy AI activity and flat financials are the norm, not the exception — and an economy-wide baseline agrees: in an April 2026 Census Bureau working paper, 66% of AI-using firms used AI solely to augment existing tasks, and 64% made no organizational changes at all.

So this assessment measures readiness the way the Ladder defines progress: by what the P&L shows. The four financial questions place you. The seven operating questions explain the placement and build the plan. Your self-guess at the start is there for one reason — most operators place themselves a stage or two above where their numbers sit, and seeing that gap is worth more than the placement itself.

The four numbers: how the assessment places you

The placement runs entirely on The Four Numbers: cost to deliver, retention, margin, and revenue per person. Each question maps to one stage signature.

Question	Why it's on the list
Has your cost to deliver a typical engagement dropped in the last two quarters?	Cost to deliver dropping is the Stage 1 signature: the first number AI moves and the easiest, because it requires no structural decision.
Are customers staying longer, or buying again more, than a year ago?	Retention is half of the Stage 2 signature: quality the customer can feel, confirmed in renewals rather than in internal dashboards.
Has margin improved in a way you can tie to how work gets delivered?	Margin is the other half of the Stage 2 signature: the efficiency finally reaching the bottom line instead of evaporating as slack.
Is revenue per person growing while headcount stays flat?	Revenue per person detaching from headcount confirms Stage 3 and nothing earlier. It only moves when the delivery model itself has changed.

Every financial question accepts three answers: it moved measurably, it's flat, or we don't track it. The third answer matters most. Answer "don't track" on two or more of the four and the assessment refuses to place you — not as a scolding, but because a firm that isn't measuring these numbers isn't at a stage. It's unplaced, and its first move is a week of bookkeeping, not an AI initiative.

The seven operating questions, and why these seven

The financial questions place you; these explain the placement and set the plan. Each one comes from a pattern I keep seeing in firms that make a stage jump — or keep failing to.

Question	Why it's on the list
Does one named person own AI at your firm?	The first of the three enablers that separate firms that make the Stage 1 to 2 jump from firms that keep trying. Ownership by everyone is ownership by no one.
Is there a short written AI policy people can actually find?	A sandbox with guardrails outperforms both silence and a restrictive policy nobody can find. People who can’t find the rule improvise one.
Is time to learn and rebuild workflows funded — on the calendar, in hours?	The Unfunded Hour: Stage 2 asks people to work a new way while still delivering the old way, and that transition cost has to be funded in hours.
Has any workflow been rebuilt around AI — steps removed or changed?	The working definition of the Stage 1 to 2 jump: the workflow gets rebuilt around AI instead of AI being bolted onto the old one.
Is your firm’s knowledge centralized where an AI tool could reach it?	Shared context before autonomous delivery: the Stage 3 precondition. A Stage 3 placement without centralized knowledge is a volume claim, not a delivery model.
Could someone reconstruct last month’s key decisions from what’s written down?	Documentation is the rate limiter on everything built later, not hygiene. You cannot automate a workflow that exists only inside someone’s head.
Have customers noticed a change in your work?	Stage 2 is confirmed from the outside. A Stage 1 firm describes faster tasks; a Stage 2 firm points to better reviews and stronger renewals.

Three of these — the named owner, the written policy, the funded time — are the enablers that separate firms that make the Stage 1 to 2 jump from firms that keep trying. They're unglamorous on purpose. None of them is a purchase, and that's the point: movement up the Ladder requires decisions that cost attention and authority, not subscription budget.

How the placement works

Gate logic, in order. No weights, no composite score — a firm has a stage, not a number.

1. The measurement gate. Two or more "don't track" answers on the four numbers → unplaced. The plan starts with instrumenting the numbers.
2. Stage 3 — AI-native. Revenue per person growing while headcount stays flat, and firm knowledge centralized where AI can reach it, and the Stage 2 signature already present. All three, because revenue-per-person growth without shared context is a volume claim, and the Ladder treats unverified volume claims as a quality debt spiral seen from the outside.
3. Stage 2 — Augmented. Retention and margin both improved — the middle pair. Exactly one of the pair moved → you're placed at Stage 1 with an explicit "on the cusp" note, because one number can move for reasons that have nothing to do with delivery.
4. Stage 1 — Enhanced. Cost to deliver dropped; the middle pair flat. The most common placement, and the most common stall.
5. Stage 0 — Assisted. Nothing moved. A starting point, not a failure — the mistake is treating it as an arrival.

The deliberate omission: nowhere does the logic ask how much AI the firm uses. A firm where every employee prompts daily can land at Stage 0, and by design. Individual use with no workflow change, no repricing, and no staffing decision is Stage 0 by definition, whatever it feels like from the inside.

What the plan is built from

The verdict ends with at most four moves, drawn from a fixed library and filtered to what you answered "no" or "partly" on. The library is the Ladder's own playbook, in Ladder order:

• Toward Stage 2: name the owner, write the boundary down, put learning time on the calendar, rebuild one workflow, and decide where the freed hours go before they evaporate. That last one is the whole difference between saving time and making money — The Production Gap catalogs how firms lose it.
• Toward Stage 3: make the work machine-readable, centralize knowledge where AI can reach it, name the verifiable units of your delivery, and pair every output claim with a flat quality metric.

Two properties are worth stating. First, the plan never recommends buying anything, because no purchase appears anywhere in the stage signatures. Second, an all-yes firm gets a short plan — the assessment doesn't pad. If you've made the moves, the remaining work is compounding them until the economics detach, and no questionnaire accelerates that.

How long until the next stage shows up in the P&L?

The verdict includes a timeline band, set by how many of the three enablers you have in place: 6–9 months with all three, 9–12 months with two, 12–18 months with one, and 18+ months — if at all with none. Those bands come from the range I observe across the network — most firms take six to eighteen months of sustained decisions for a stage transition to become visible in the financials, and firms missing all three enablers routinely never make the trip.

The bands are the model's one judgment call, and I'd rather state that than imply precision the data doesn't carry. What the data does support: the enablers predict movement better than budget does, because the binding constraint in most firms isn't the subscription line. It's hours — people who would use the tools and haven't been given time to learn them.

What this diagnostic can't tell you

Three honest limits. The answers are self-reported, and the questions the assessment most wants answered honestly — did margin really move because of delivery? — are the ones easiest to answer generously. If you want the harder version, pull the last two quarters of financials before you start, and answer from the pages, not from memory.

It can't verify quality. A Stage 2 or Stage 3 placement here means the signature is present, not that the claim would survive an audit. The real test is the pairing: more delivered at the same rework rate, faster turnaround at the same escalation quality, measured across quarters.

And it can't make the decision. The jump from Stage 1 to Stage 2 — the hard one — starts when someone with authority decides the workflow itself gets rebuilt. The assessment can tell you that decision hasn't been made. It can't make it for you.

---

The AI Readiness Assessment runs on the three Proofwork models: The Delivery Model Ladder for the stages, The Four Numbers for the measurement, and The Production Gap for the ways firms get stuck between stages. Nothing you enter leaves your browser; the shareable link encodes your answers and nothing else. For more on how service firms rebuild delivery around AI, subscribe to the newsletter.