What a Free AI Readiness Assessment Can and Cannot Tell You

A free AI readiness assessment can tell you whether a formal reading is worth commissioning. It cannot tell you where your business stands. Ours asks a short set of self-report questions, scores them in your browser, and returns a plain-language band rather than a number — a starting sense, deliberately not a measurement.

The difference is not one of degree, and it is worth spelling out. That is an awkward thing to publish on the same site that offers the assessment, so it is worth saying who is writing. Blue Horizon Labs built the self-serve readiness self-assessment, and it exists in part because people who take it sometimes go on to commission a diagnostic. Read the account below with that in mind. The reason to write it anyway is that a lead-generation instrument nobody has described honestly is the kind of thing that quietly makes a market worse — and the limits below are the same ones we would state on a call.

What it actually reads

Ten questions, four dimensions, three bands at the end. The four dimensions are not invented for the tool. They are four of the five dimensions the full diagnostic scores, used verbatim: strategic clarity, operational efficiency, technology enablement, and organizational capability. The vocabulary is the same on purpose, so that a person who later sees a real reading is looking at familiar categories rather than a second, unrelated model.

Each question is answered on a three-point scale that runs from the operation depending on heroics to the operation compounding on its own. The scoring happens in the page. Nothing is sent anywhere: no request leaves the browser, nothing is written to storage, and no analytics event fires when you answer. We do not know your result and cannot look it up. That is a design decision rather than a privacy posture — an assessment whose answers are being collected is an intake form, and people answer intake forms strategically.

The fifth dimension, and why it is missing

The full Performance Index reads five dimensions. The self-assessment reads four. The one deliberately left out is Revenue Architecture — how predictably and durably revenue compounds — and the omission is the most informative thing about the instrument.

Revenue Architecture is the least legible dimension to an owner from the inside without financial instrumentation. Asking about it in a self-reading would not produce an honest answer; it would produce a guess, delivered with the confidence of a self-report, and that guess would then be scored as if it were an observation. A question that reliably returns a guess makes an instrument worse, not more complete. So it is absent, and the reading it produces is narrower as a result.

That is the general shape of what a free assessment can and cannot do. It can ask about things you can see from where you sit. It cannot ask about the things you would need an instrument to see, which are frequently the things that matter.

Three bands, and no number

The result is one of three bands, not a score. There is no "73 out of 100" at the end, and there never was — the underlying total is a private threshold, never displayed, because a two-digit figure derived from ten self-reported answers would carry a precision the input cannot support. A number invites you to compare yourself to someone else's number. A band invites you to read a paragraph.

Each band says three things: what it means, what it does not mean, and one honest next step.

Foundations first means the structure underneath the operation is not settled enough for AI to help yet — put AI on top of it today and it would mostly automate the parts that already are not working, and make them harder to see. It does not mean the business is behind. Most owner-led companies read here at first.

Ready for a reading means there is real structure to work with, enough that a diagnostic could find specific places where the work would pay rather than guessing at them. It does not mean you are ready to deploy AI across the board. Readiness is uneven; some parts are solid and others are not.

Built to compound means the operation already runs on documented, owned systems that connect. And the band says the most useful sentence in the whole instrument: it does not mean the work is done, or that this self-reading has confirmed it — a structured self-assessment can be generous with itself, and a scored diagnostic is what actually tests the claim.

That last line is the honest summary of the entire category. A self-assessment is a conversation with yourself about your own business, conducted using categories somebody else supplied.

What it cannot tell you at all

Three things, none of which is a limitation of our version specifically.

It cannot tell you the gap between how the business runs and how you believe it runs. That gap is the single most valuable output of a real diagnostic, and it is structurally invisible to a self-report, because the person answering is the person holding the belief.

It cannot find the issue you have not named. The questions can only cover ground someone anticipated. A diagnostic runs twelve scorecards across five dimensions over two to four weeks and produces an issue tree that separates the top three structural issues from the symptoms people report — and the reason that exercise is worth its cost is precisely that the top three are usually not the three the owner would have listed.

It cannot be wrong in a way you would notice. A broken instrument that returns a plausible band looks exactly like a working one. There is no error message. The only way to know whether a self-reading was accurate is to take a real measurement afterwards, which is the same reason a page like this one should exist.

We also do not mark the assessment up as something it is not. The page carries no quiz structured data, because the scoring is qualitative self-guidance rather than a graded quiz, and marking it up as a quiz would overstate what it is to every machine that reads the page.

When the free version is enough

Often, and more often than a firm selling diagnostics would like to admit.

If the assessment returns Foundations first and you recognise the description, you have your answer and you do not need to pay anyone for it. The next move is structural and it is yours: get one core workflow documented and owned. That moves readiness further than any model would, and it costs a week of attention rather than a fee.

If you are below the band this work is built for — roughly $5M to $50M in revenue, usually owner-led, where growth has outrun the structure underneath it — a formal reading is probably premature in any case. Under that, the owner is still close enough to every function that informal structure is the correct structure, and the honest instrument is a hard conversation and a whiteboard.

And even inside the band, the reading is frequently the whole engagement. About six diagnostics in ten end with a documented score and a set of findings the owner executes independently. If a business needed a reading and now has one, it does not need a lab past that point.

The free version earns its place by answering exactly one question: is a real measurement worth commissioning? That is a genuinely useful question, and it is the only one an instrument this light can answer. Anything more confident than that — from us or from anyone else — is a marketing asset wearing an instrument's clothes.

If you want the scored version, the diagnostic is the instrument that produces it — and what it actually measures is a longer answer than a band can carry.