AI Readiness Checklist: 24 Questions to Ask Before You Invest in AI

This is a self-diagnosis tool. It gives you 24 questions, grouped four to a dimension across the six dimensions that predict whether AI works inside an organization: Data Foundation, Team & Culture, Technology, Processes, Strategy, and AI Adoption. For each question you compare your honest answer to a "ready" pattern and a "not ready" pattern, and you note which side you fall on.

It is not a pass-or-fail test, and there is no magic number of "ready" answers that means you are cleared to spend. The point is to surface, in about twenty minutes, the specific prerequisites you have and the specific ones you are missing, so the AI use case you choose matches the readiness you actually have, rather than discovering the mismatch six to eighteen months and a budget later.

This is also the static version of the free AI Opportunity Score. The questions here are the same ones the quiz scores for you. If you would rather have your answers turned into a 0-100 result with a per-dimension breakdown and a radar chart, take the quiz; if you want to think through each question carefully and see the reasoning behind it, read on.

Answer for your organization as it is today, not as you intend it to be after a project you have not started. The two most common ways these answers go wrong are over-rating your data ("we have a CRM, so our data is good") and under-rating your culture ("I'm excited about AI, so the team must be too"). When in doubt, answer one notch more conservatively than your instinct.

For a sharper read, have three people answer independently, a senior leader who owns strategy and budget, a technical lead who knows your systems, and an operations lead who knows how the work actually gets done, then compare. Where they disagree on the same question, that disagreement is the finding: it usually means the dimension is in transition or inconsistent across teams.

Keep it simple: for each of the six dimensions, count how many of its four questions landed on the "not ready" side. A dimension with three or four "not ready" answers is a blocking dimension, it will cap any AI use case that depends on it, no matter how strong your other dimensions are. A dimension with zero or one is a strength you can build a first use case around. We come back to how to read the full pattern in section 8.

If you would rather not tally by hand, the AI Opportunity Score does the weighting for you and plots the result as a radar chart, which makes the gap between your strongest and weakest dimension obvious at a glance, and that variance is often the most important thing the exercise reveals.

Ready: The outcomes you would want AI to predict or assist with are already being recorded somewhere, sales outcomes, support resolutions, production results, customer behavior. The raw material exists, even if it is messy.

Not ready: The decisions you want to improve happen in people's heads, in email threads, or in spreadsheets that get overwritten, and there is no durable record of inputs and outcomes to learn from.

Why it matters: AI cannot find a pattern in data you never recorded. If the history does not exist, no tool can recover it, and any data-dependent use case stalls at the very first step while you start collecting from scratch.

Ready: Key fields are populated reliably, categories and labels are applied consistently, duplicates are rare, and the people who use the data already treat its outputs as trustworthy for decisions.

Not ready: Records routinely miss key fields, the same thing is labelled three different ways, and staff keep private side-spreadsheets because they do not trust the system of record.

Why it matters: Models inherit the quality of their inputs. Train on inconsistent data and you get confident, wrong outputs, which is worse than no model, because a wrong answer that looks authoritative erodes trust and gets the whole initiative cancelled.

Ready: The data that matters can be brought together, through a warehouse, connected systems, or reliable exports, so a single use case can draw on customer, product, and operational data at once.

Not ready: Customer data lives in the CRM, usage data in a separate product tool, and finance data in yet another system, with no connection between them and no realistic way to join them.

Why it matters: The most common first use cases, churn prediction, lead scoring, demand forecasting, require signals from more than one system. If your data cannot be joined, those use cases are blocked until integration work happens, and that work is usually the longest part of the project.

Ready: Core terms like "active customer," "qualified lead," or "defect" mean the same thing across teams, someone owns data quality, and there are rules for who can see and change what.

Not ready: Two departments define "active customer" differently, nobody owns the data, and there are no access or privacy controls beyond whatever each system happens to enforce.

Why it matters: AI will faithfully reproduce your definitional inconsistencies at scale, and ungoverned data becomes a compliance liability the moment it flows into an AI tool. Governance is what lets you trust an output and defend how it was produced.

Ready: At least a few people already use AI tools in their daily work and can tell you, from experience, where they help and where they fall short.

Not ready: AI is something the organization has read about but never used, and no one internally can judge a vendor claim or evaluate an output critically.

Why it matters: A skills gap is the most fixable readiness problem, most operational use cases can be taught in a matter of weeks. But with zero hands-on experience you cannot tell a realistic AI plan from a fantasy one, which is exactly when budgets get wasted on the wrong tool.

Ready: Senior leadership has said AI matters, backed it with budget, and is willing to spend its own attention on it, not just delegate it downward and hope.

Not ready: AI is a buzzword in a strategy deck with no funding attached, or it is treated as purely an IT project that leadership expects to happen without their involvement.

Why it matters: Adoption requires people to change how they work, and people change behavior when leadership clearly and repeatedly signals that it matters. Without visible executive commitment, even technically successful pilots quietly fail to scale.

Ready: The people who would actually use AI in their day are curious or at least willing, and do not see it primarily as a threat to their jobs or their craft.

Not ready: Front-line staff or senior practitioners actively resist automation, block tools, or quietly refuse to change a workflow they consider theirs.

Why it matters: A tool nobody adopts produces no value regardless of how good it is. Cultural resistance is the slowest readiness gap to close, it can take quarters of consistent messaging and demonstrated wins, so it is worth knowing about before you commit, not after.

Ready: You have rolled out a major system, a CRM, an ERP, a new operating process, and made it stick, which means you have the change-management muscle AI adoption needs.

Not ready: Past rollouts stalled, were abandoned, or are remembered as painful failures, and there is no track record of getting the organization to adopt something new.

Why it matters: Deploying AI is a change-management problem at least as much as a technology problem. Organizations that have done a hard rollout before know how to retrain people, redesign workflows, and measure the result, the same muscles AI requires.

Ready: Your main systems are cloud-based or API-connected, so data can flow to an AI service and results can flow back into the tools your team already uses.

Not ready: Critical systems are on-premise monoliths or legacy applications with no API layer, where getting data in or out means manual exports or custom engineering.

Why it matters: AI delivers value when it is wired into the flow of work, not when it sits in a separate window someone has to copy and paste into. Without connectivity, every AI interaction becomes manual, adoption collapses, and the deployment timeline stretches by months.

Ready: You already run on a major cloud platform or have the integration capability to connect to AI services, so standing up a working pipeline is a matter of weeks, not a re-platforming project.

Not ready: Connecting anything to an external AI service would require infrastructure you do not have, and there is no clear path from your current stack to a live model.

Why it matters: This is the difference between an AI project that ships in a quarter and one that needs a foundational infrastructure investment first. Knowing which situation you are in keeps your timeline and budget honest.

Ready: You can control and audit which data is allowed to reach an AI tool, who can use which AI services, and how sensitive information is handled before it ever leaves your perimeter.

Not ready: There are no controls on what staff can paste into public AI tools, no visibility into which services are in use, and no answer to "where does our data go when someone uses this."

Why it matters: AI introduces data-exposure and access risks that traditional security tooling does not automatically cover. Without governance you are one careless prompt away from leaking confidential data, and you cannot deploy AI in any regulated context safely.

Ready: You have a way to check that an AI tool keeps producing good results after launch, sampling outputs, tracking error rates, or catching when quality slips.

Not ready: Once a tool is live you have no way to tell whether it is still working well, and you would only find out it had degraded when something visibly broke.

Why it matters: AI outputs drift as the world and your data change. Without monitoring you will trust a model that has quietly stopped being reliable, and the failure surfaces as a customer or compliance problem rather than a dashboard alert.

Ready: The steps, inputs, outputs, and exceptions of your target process are written down clearly enough that a new hire could follow them.

Not ready: The process lives only in the heads of experienced staff, and asking three of them how it works would produce three different answers.

Why it matters: You cannot configure or train AI for a process nobody has defined. Documentation is the prerequisite step, and if it does not exist, that mapping work is the real first task, before any tool is involved.

Ready: The work is done the same way regardless of who does it or which shift it falls on, against shared standards for what "done" and "good" mean.

Not ready: Every person does it their own way, the criteria for a good result shift with whoever is judging, and outcomes are inconsistent for reasons no one can pin down.

Why it matters: AI learns the pattern in how a process is performed. If there is no consistent pattern, only individual variation, there is nothing stable to learn, and you have to standardize the process before automating it.

Ready: You have a baseline, time taken, error rate, throughput, cost per unit, so you could prove whether AI made the process better or worse.

Not ready: There are no numbers on the current process, so "did AI help" would be a matter of opinion rather than evidence.

Why it matters: Without a baseline you cannot justify continued AI investment, you cannot tell success from expensive failure, and the project becomes impossible to defend the moment someone asks for its return.

Ready: You have the project-management capability to change how work flows once AI is in the loop, reassign steps, define when a human reviews the AI, retrain the people affected.

Not ready: Workflows are rigid or informal, there is no one who owns redesigning them, and "the AI suggests it but we still do it the old way" is the likely outcome.

Why it matters: An AI output that nobody acts on changes nothing. Value comes from rebuilding the workflow around the output and defining the human checkpoints, which takes process ownership, not just a model.

Ready: You can name the business outcome AI is meant to move, reduce churn, cut handling time, increase conversion, not just "do something with AI."

Not ready: The goal is to "adopt AI" or "not fall behind," with no connection to a number the business actually cares about.

Why it matters: AI tied to a vague ambition has no success criteria, so it can never clearly succeed. Use cases anchored to a specific objective get prioritized, measured, and funded; the rest drift.

Ready: Leadership has chosen a small number of priority use cases and explicitly deprioritized the rest, so effort is concentrated rather than scattered.

Not ready: Everyone has a different pet idea, a dozen possibilities are "being explored," and nothing has been chosen over anything else.

Why it matters: Spreading effort across many simultaneous use cases dilutes implementation quality, the most common reason otherwise-ready organizations stall. Agreement on what comes first is what turns interest into delivery.

Ready: Real budget is set aside for tools, integration, and the people time to implement and adopt, enough to take one use case all the way to production.

Not ready: The plan relies entirely on free tiers and spare time, with no funding for the integration and change-management work that production actually requires.

Why it matters: The cost of an AI tool is usually the smallest line item; integration, data work, and adoption cost more. Underfunded AI stays stuck at the pilot stage because there is never enough to cross into production.

Ready: One named person is responsible for coordinating AI work, tracking what is in use, and reporting to leadership, even if it is not their full-time role.

Not ready: AI is "everyone's job," which means it is no one's, and there is no single point of accountability for whether anything ships.

Why it matters: Without ownership, nothing scales. Pilots end and quietly die because no one had the mandate to make the prioritization, resourcing, and organizational decisions that production requires.

Ready: You have run at least one AI initiative through to a real conclusion and learned firsthand where the friction lives, in data, integration, or adoption.

Not ready: You have never run an AI project, so every implementation challenge ahead is one you will be meeting for the first time.

Why it matters: The first deployment is always the hardest. Having done one before means the team has already absorbed the surprises that derail first-timers, this does not disqualify newcomers, but it tells you to plan for a longer learning curve.

Ready: At least one AI effort produced a result you could measure, time saved, errors reduced, throughput up, not just a demo that impressed people in a meeting.

Not ready: Pilots were technically interesting but their impact was never quantified, so you cannot say whether they actually helped the business.

Why it matters: A measured outcome proves you can connect AI to business value, which is the whole point. Organizations that only run unmeasured demos tend to accumulate a pilot graveyard rather than a track record.

Ready: An AI tool is genuinely embedded in how work gets done, a writing assistant most of the team uses, an automated classifier in the pipeline, a forecast that feeds weekly planning.

Not ready: AI has only ever existed as a trial or experiment, never crossing into something people depend on day to day.

Why it matters: Production use is the only stage at which AI investment converts into business value, and reaching it proves you cleared every barrier between idea and daily reliance. That demonstrated capability transfers to harder projects.

Ready: Your team can speak concretely about what went wrong and what they would do differently, the gap between an AI vendor's pitch and the messy reality of making it work.

Not ready: Any view of AI implementation is theoretical, drawn from articles and demos rather than the experience of having shipped something.

Why it matters: Earned lessons are the difference between a realistic plan and an optimistic one. Teams with scars set sensible timelines and budgets; teams without them tend to under-estimate every difficulty and over-promise the result.

The distribution of your "not ready" answers matters more than their sum. An organization with four "not ready" answers all concentrated in Data Foundation is in a very different position from one with four spread one-per-dimension. The first has a single, identifiable blocker to fix; the second has a thin foundation everywhere.

A simple rule: a dimension with three or four "not ready" answers is a blocking dimension, one with two is a watch dimension, and one with zero or one is a strength. Write down which dimensions fall into each bucket, that list is the actual output of this exercise.

The strategic principle behind the whole checklist: your weakest dimension caps your most AI-intensive use cases. You can work around a weak dimension by choosing use cases that do not depend on it, a thin Data Foundation does not stop you using generative writing assistance, but you cannot ignore it indefinitely if you intend to build AI capability at scale.

So the move is not "raise the easiest dimension to make the average look better." It is "fix the blocking dimension, even though it is usually the hardest and slowest to move," or "pick a use case that routes around it for now." Either is a deliberate choice; drifting into a data-dependent project on a weak Data Foundation is not.

If Data Foundation is your blocker, a generative use case, AI-assisted writing, summarization, research, coding help, is the realistic starting point, because it draws on a model's general knowledge rather than your data. If Strategy is your blocker, no use case is safe until leadership picks one, funds it, and names an owner; the fix is a meeting, not a tool. If Processes is your blocker, the first deliverable is documentation and standardization, before any model is configured.

This is the difference between this static checklist and a generic "are you AI-ready" quiz: it does not just give you a verdict, it tells you which use case fits your current shape and which prerequisite to close first.

Hand-marking 24 questions tells you the shape of your readiness. The free AI Opportunity Score turns that shape into a number: a weighted 0-100 overall result, a score for each of the six dimensions, and a radar chart that makes the variance between your strongest and weakest dimension visible instantly, no signup, about three minutes.

If you want the reasoning behind the framework, what each dimension measures, how to interpret the four maturity bands (Beginning, Developing, Advanced, Leading), what to do at each band, and how this compares to the Gartner, McKinsey, MIT, and Deloitte frameworks, the cornerstone AI readiness assessment guide covers all of it. Use this checklist to diagnose; use the guide to understand what your diagnosis means.

An AI readiness checklist is a structured set of yes-or-no questions that tells you whether your organization has the prerequisites to adopt AI successfully. This one uses 24 questions, four under each of six dimensions: Data Foundation, Team & Culture, Technology, Processes, Strategy, and AI Adoption. For each question you compare your honest answer to a "ready" pattern and a "not ready" pattern. It is the static version of the free AI Opportunity Score, which scores the same questions automatically and returns a 0-100 result with a radar chart.

Work through all 24 questions and count how many land on the "not ready" side in each dimension. There is no single pass mark. The signal is the distribution: a dimension with three or four "not ready" answers is a blocking dimension that caps any AI use case relying on it, no matter how strong your other dimensions are. Most organizations are ready for some narrow use case right away and not ready for broad deployment until they close one or two specific gaps.

Twenty-four, divided evenly across the six readiness dimensions, four each for Data Foundation, Team & Culture, Technology, Processes, Strategy, and AI Adoption. Four per dimension is enough to cover the sub-factors that predict outcomes without turning the exercise into a multi-week audit.

For any use case that learns from your own historical data, the most decisive question is Q3, whether that data is connected and accessible in one place. Data Foundation is the dimension most correlated with AI project success, and a fragmented-data answer blocks the most common first use cases. For generative use cases that rely on a third-party model's knowledge rather than your data, the most important question shifts to Strategy: whether you have agreed on a specific, owned, funded use case.

It does not mean you cannot use AI. It means a specific prerequisite is missing for a specific class of use case. A weak Data Foundation blocks use cases that learn from your data, but not generative writing or coding help that rely on a model's general knowledge. The whole purpose of the checklist is to match the use case you choose to the readiness you actually have, rather than discovering the mismatch after the budget is spent.

Yes, this page is the static, self-marked version of the same 24 questions. The free AI Opportunity Score scores them for you: a 0-100 overall result, a per-dimension breakdown, and a radar chart that makes the variance between your strongest and weakest dimension obvious at a glance. No signup, about three minutes.

The most useful single respondent is someone who can see across strategy, technology, and operations, typically a CEO, COO, CTO, or head of transformation at a small or mid-size company. For a more accurate picture, have a senior leader, a technical lead, and an operations lead each answer independently and then compare. Disagreement on the same question is itself a signal: it usually means that part of the organization is in transition or inconsistent across teams.

Every six months for organizations actively investing in AI readiness, and annually for those in early planning. Redo it sooner after any event that would change an answer: a data integration project, an AI-focused hire, a successful or failed deployment, or a regulatory change affecting your sector. Because the questions are stable, re-answering them over time gives you a clean before-and-after on each dimension.

This checklist is for self-diagnosis: 24 concrete questions you can answer right now to see where you stand. The full AI readiness assessment guide explains the concept end to end, what each dimension measures and why, how to interpret scores and dimensional variance, what to do at each of the four maturity bands, and how the framework compares to Gartner, McKinsey, MIT, and Deloitte. Use this checklist to diagnose; use the guide to understand what your diagnosis means and what to do next.