Readiness Is a Prerequisite List, Not a Score

Artificial intelligence readiness is not a feeling. It is the ability to name the function to improve, the data that function produces, and the person who owns the outcome. Anything less is optimism dressed as planning.

Most readiness assessments fail because they produce a score rather than a list. A score obscures which prerequisite is missing. Lists make gaps visible and assign them to people.

The anti-pattern is the maturity model

A familiar pattern runs through companies preparing for artificial intelligence adoption. Consultants arrive with frameworks that rank readiness on a scale. The company scores itself, discovers it is at the middle, and concludes that it is almost ready.

That conclusion is dangerous because a middle score means nothing specific. Every prerequisite could be partially met, or one could be entirely absent while the others are complete. The score does not distinguish those states.

Maturity models have a second flaw. They treat readiness as a property of the whole organization rather than of a specific function. Artificial intelligence does not land on an organization. It lands on a process, and that process has its own prerequisites.

Do not score, list

A calmer response to readiness pressure begins with a diagnostic pause. Before any vendor conversation, the company needs a written list of prerequisites for the one function it intends to improve.

That list has three categories. Data prerequisites ask whether inputs exist in retrievable form and whether they are clean enough to produce reliable output. Ownership prerequisites ask whether one person is named as the decision maker for whatever the system recommends.

Fallback prerequisites ask what happens when the system is wrong and whether a human process exists to catch the error. Any function missing an item in any category is not ready. Not partially ready. Simply not ready, and the specific missing item is the only thing that matters.

Theory of constraints supports this discipline by focusing improvement on the one function that limits throughput. A VRIO analysis adds the lens of whether the data is valuable, rare, inimitable, and organized. Functions whose data fails the VRIO test are not candidates for artificial intelligence intervention. They are candidates for data cleanup.

The systemic fix is a prerequisite audit

A serious position on ai readiness treats it as a pass-fail audit rather than as a numeric assessment. The audit asks binary questions about a specific function. The result is a list of gaps, each assigned to a person with a deadline.

Step one is function selection. Not organization-wide readiness. One function, the one that constrains throughput, documented and owned and measured. Artificial intelligence applied anywhere else is local motion.

Step two is data inventory. For that function, what data exists, where it lives, who maintains it, and how often it is wrong. Data with no owner is not an asset. It is a liability.

Step three is decision mapping. For each output the system might produce, who decides whether to act on it and what happens if they disagree. A recommendation with no decision owner is noise.

Step four is fallback design. When the system produces an outlier, who catches it, how, and by when. Without a fallback, the organization is betting its operations on the accuracy of a tool it has not yet tested.

A RACI grid is useful across all four steps, because most readiness gaps turn out to be ownership gaps wearing technical clothing. Data may exist yet lack a maintainer. Decisions may matter yet lack a named owner. Fallbacks may be obvious yet remain unpracticed.

Firms that complete this audit before contacting vendors discover that their requirement becomes specific enough to disqualify most of the market. That specificity saves months of evaluation time and prevents the common trap of adapting the process to fit the tool.

Why this is an intellectual discipline question

Readiness is not a state of inspiration. It is a state of documentation. A company that cannot write down its process, its data sources, and its decision owners is not almost ready. It is missing the foundation that any system requires.

The discipline here is citational logic. Every claim about readiness must point to a specific document, a specific owner, and a specific test. Claims without those three are not claims. They are hopes.

That discipline protects the company from adoption theater that consumes budget and patience. A prerequisite audit takes longer than a maturity survey and it produces something the survey cannot. A named gap, a named owner, and a named deadline.

What this looks like in practice

Consider a mid-market manufacturer that wanted to forecast demand more accurately. Maturity models said the company was moderately ready. Prerequisite audits said the historical data was stored in three different systems with inconsistent categorization and no owner.

The data existed. It was not retrievable in a form that any system could use without extensive reconciliation. The audit exposed that gap in a single afternoon. The maturity model had hidden it behind an average.

Cleaning the data took months, not because the work was complex, but because the categorization had to be agreed across departments that had defined it differently for years. That agreement was the real readiness work. Once it was done, the system installation was straightforward.

Organizations that run prerequisite audits before vendor selection report a consistent effect. Their vendor conversations are shorter and more specific, because the requirement is clear enough that most candidates disqualify themselves. The shortlist that remains is short for the right reason.

Why this protects human capital

Companies that install artificial intelligence on unreadied processes do not merely waste budget. They force their people to compensate for the gap between what the system promises and what the process can deliver. That compensation is invisible, exhausting, and eventually blamed on the people rather than on the readiness failure.

The prerequisite audit is a form of care because it refuses to ask people to bridge a gap that documentation should have closed. It insists on process clarity before process acceleration, which means the people inside the process are protected from being the fallback.

That protection is the moral core of operational discipline. Systems should serve the people who run them. People should not serve as the error correction for systems that were installed before the process was ready.

What compounds

Firms that treat readiness as a prerequisite list accumulate something no maturity score delivers. They develop the habit of binary assessment, which makes every subsequent technology decision faster and more honest. A company that can audit one function can audit any function.

That habit spreads. The same discipline applied to artificial intelligence readiness can be applied to any operational change, because the universal questions are the same. Does the data exist and is it owned. Is the decision mapped and is the fallback designed.

A balanced scorecard is useful here because it forces the company to state what improvement means before claiming any system delivered it. Readiness is not the score. The score is what the company wants to move, and readiness is the list of prerequisites that must be true before the attempt is worth making.

Shared understanding across departments is what makes the audit stick. When finance and operations agree on categorization, the data set becomes usable for both. That alignment is a collaboration outcome that outlasts any single project.

Every function a company can describe with a documented process, a named owner, and a measured outcome is a function ready for improvement of any kind. Every function that lacks one of those three is a function where any intervention, technological or otherwise, will struggle to find footing.

Frequently Asked Questions

Why is a prerequisite list better than a maturity score?

Lists name the exact gap and assign it to a person. Scores average across prerequisites and hide which one is missing. A company with a middle score could be missing one critical item or all of them partially, and the score does not say which.

What are the three categories of readiness?

Data prerequisites ask whether inputs exist in retrievable form. Ownership prerequisites ask whether one person owns the decision the system informs. Fallback prerequisites ask what happens when the system is wrong and whether a human process catches the error.

Can a company be ready for artificial intelligence organization-wide?

No. Artificial intelligence lands on a specific process, and that process has its own prerequisites. A function with clean data and a named owner may sit beside a function with neither. Organization-wide scores obscure those differences.

What is the most common readiness gap?

Data that exists but has no owner. Multiple systems hold overlapping information, categories are inconsistent across departments, and nobody is responsible for keeping the data current. This gap is organizational rather than technical.

How long should a prerequisite audit take?

One focused afternoon for a single function, assuming the right people are in the room. The audit is not complex. It is simply a set of binary questions that most organizations have never been asked in sequence.

When does outside help make sense?

When the audit keeps being postponed because everyone capable of conducting it is inside the function being examined. An outside operator brings the template and the distance needed to see assumptions that insiders no longer notice.