VI
ENAI Readiness Assessment: A 20-Point Checklist Before You Invest (2026)
Free AI readiness assessment: score your business 0-20 across data, process, people, tech, and budget before you invest. Every stat sourced.
Date Published:
By
MONA Global
Direct answer: An AI readiness assessment scores your business across five areas: data, process, people, tech stack, and budget/governance, before you spend on AI. Use the 20-point checklist below: 0–8 means fix fundamentals first, 9–14 means you can run one narrow pilot, and 15–20 means you're ready to scale AI investment with a real roadmap.
What an AI Readiness Assessment Is
An AI readiness assessment is a structured check of whether your data, processes, people, systems, and budget can actually support an AI initiative before you fund one. It doesn't ask "should we use AI" in the abstract. It scores specific, answerable yes/no questions across five areas, then converts that score into a go/no-go decision and a prioritized list of what to fix first.
Most companies skip this step and jump straight to a vendor demo or a pilot. That's how six-figure budgets end up funding a chatbot nobody uses or a prediction model trained on data too thin to predict anything. The checklist below is the free, self-serve version of the same audit MONA runs as a paid engagement on our AI consulting practice. Score yourself first, and only bring in outside help for the gaps you can't close alone.
How Ready the Average Business Is for AI
Most businesses have started using AI somewhere, but very few have the foundation to make it pay off. Adoption has outpaced readiness, which is exactly why so many AI initiatives stall after the pilot.
Metric | Figure | Source |
|---|---|---|
Organizations using AI in at least one business function | 88% (up from 78% a year earlier) | |
Organizations that qualify as fully "AI-ready" (Cisco's top "Pacesetter" tier) | 13% | |
AI projects that fail to deliver their intended business value | 80%+ (roughly double the failure rate of non-AI IT projects) | RAND Corporation — Why AI Projects Fail and How They Can Succeed |
Generative AI pilots that deliver no measurable P&L impact | 95% | |
AI projects Gartner projects will be abandoned by end of 2026 for lack of AI-ready data | 60% |
The gap between "we use AI" (88%) and "we're actually ready" (13%) is the whole story. RAND's interviews with 65 data scientists and engineers, and MIT's review of 300 enterprise AI deployments, point to the same root causes over and over: no shared definition of success, weak data foundations, no accountable owner, and leadership sponsorship that evaporates the moment the pilot stops being new. None of those are technology problems, which is exactly why a readiness assessment, not a better model, is the fix.
The 20-Point AI Readiness Checklist
The 20-Point AI Readiness Checklist (AI-generated illustration)
Answer yes or no to each question. Be honest: "we're working on it" counts as no. Score 1 point per "yes."
Data Readiness
# | Question | Why it matters |
|---|---|---|
1 | Do you have one identifiable source of truth for the data this use case needs — not the same fact living in three disconnected spreadsheets? | Fragmented data is the most common reason Gartner expects 60% of AI projects to be abandoned by end of 2026. |
2 | Can that data be pulled programmatically (API, database export), instead of someone exporting a CSV by hand every time? | Manual pipelines work in a demo and break the first week in production. |
3 | Do you have enough historical volume and time-depth to train or ground a model, not just a few thin months of logs? | Forecasting and scoring models need history; a knowledge-base chatbot needs documents broad enough to answer real questions without inventing answers. |
4 | Has anyone audited this data for accuracy, duplicates, and staleness in the last 12 months? | Dirty training data doesn't produce a cautious model — it produces a confidently wrong one, at scale. |
Process Readiness
# | Question | Why it matters |
|---|---|---|
5 | Have you measured — not estimated — the hours or errors the target process actually costs today? | You can't prove ROI on a process you never baselined; "it feels slow" isn't a business case. |
6 | Is the process documented well enough that someone unfamiliar with it could follow the steps? | If the process only lives in one person's head, there's nothing consistent for an AI system to learn or replace. |
7 | Is there one specific business metric (hours saved, dollars, error rate) this initiative is meant to move? | RAND's research found "misaligned purpose" — no shared definition of success — among the top causes of AI project failure. |
8 | Does an existing review or QA step exist that AI output could plug into, rather than needing oversight invented from scratch? | AI without a review step is how an error reaches a customer before anyone notices. |
People Readiness
# | Question | Why it matters |
|---|---|---|
9 | Is there one named person accountable for whether this AI initiative succeeds — not a committee, not "IT"? | Diffuse ownership is a graceful way to guarantee nobody owns the failure either. |
10 | Will the people who do this work daily actually use the tool, or were they excluded from deciding what "AI here" even means? | MIT's research found the winning 5% of GenAI deployments were the ones built with frontline workflows, not imposed on them. |
11 | Do you have someone assigned to review AI outputs and catch errors before they reach a customer or a decision? | Human-in-the-loop isn't a nice-to-have for anything with financial, legal, or customer-facing stakes — it's the difference between a mistake and an incident. |
12 | Has a leader with real budget authority committed past the pilot phase? | Fading executive sponsorship is one of the leading causes RAND identifies for AI projects that quietly die after a promising demo. |
Tech Stack Readiness
# | Question | Why it matters |
|---|---|---|
13 | Can the systems this use case touches be integrated via API, or are they locked inside legacy software with no way out? | An AI system that can't reach your real data is a very expensive demo. |
14 | Do you have the cloud, compute, and security foundation to run this reliably at the volume you'll actually need? | A model that works on 50 test records and falls over at 50,000 live ones isn't ready — it's untested. |
15 | Have you deliberately chosen build vs. buy vs. orchestrate for this use case? | The right answer is sometimes a $50/month subscription; defaulting to "build custom" because a vendor pitched it is how budgets balloon. |
16 | Is there a plan to monitor, version, and update the system after launch? | Models drift and data changes; a roadmap that stops at "go live" is a roadmap for silent decay. |
Budget & Governance Readiness
# | Question | Why it matters |
|---|---|---|
17 | Does your budget include data cleanup, integration, and ongoing monitoring — not just the model or subscription line item? | The license fee is usually the smallest cost in the project; teams that budget only for it run out of money mid-build. |
18 | Do you have a written policy on what data may or may not touch third-party AI APIs? | Skipping this until after a vendor call is how sensitive customer or financial data ends up somewhere it shouldn't. |
19 | Have you scoped one narrow, specific use case, rather than an open-ended "AI transformation" mandate? | Open-ended mandates never hit a finish line to measure against — narrow use cases do. |
20 | Do you have a fixed review window (e.g., 90 days) and a metric that decides scale, adjust, or kill? | Without a decision point, pilots don't get killed when they should — they just quietly keep costing money. |
How to Score Your AI Readiness Assessment
Add up your "yes" answers out of 20.
Score | Readiness level | What it means |
|---|---|---|
0–8 | Not ready | Foundational gaps in data, process, or ownership will sink most AI spend right now. Fix the fundamentals before funding a pilot. |
9–14 | Pilot-ready | You have enough in place to run one narrow, well-scoped pilot — but not enough to fund a company-wide AI program yet. |
15–20 | Scale-ready | Your foundation can support a real roadmap: multiple prioritized use cases, sequenced investment, and governance built in from day one. |
5 Signs You Shouldn't Invest in AI Yet
5 Signs You Shouldn't Invest in AI Yet (AI-generated illustration)
Content that never says "don't buy" doesn't get trusted or cited. If any of these are true, the honest move is to fix them before spending on AI, not after:
- You can't name the specific metric AI would move. "We should do something with AI" is a goal in search of a problem, not a business case.
- Nobody owns the data, and cleaning it isn't already on someone's roadmap. An AI initiative bolted onto ungoverned data is a data-cleanup project wearing an AI costume.
- Your best process for the target task is undocumented, one person's tribal knowledge. There's nothing consistent yet for a system to learn, automate, or be measured against.
- Leadership wants a demo for a board slide, not a system with a maintenance budget. A demo that never gets a second quarter of funding was never really a project.
- There's no human-in-the-loop plan, and being confidently wrong here is expensive. Anywhere errors touch money, compliance, or a customer relationship, skipping review isn't fast. It's a future incident.
None of these are permanent. They're usually a 30–90 day fix, which is a far better use of that time than a pilot that was doomed before it started.
What to Do Next, Based on Your Score
- Scored 0–8? Don't hire an AI vendor yet. Start with the process and data gaps the checklist flagged. Often that means an operations-level audit before anything AI-specific. Our automation consulting practice is built for exactly this stage: mapping which processes are worth fixing first and measuring the hours they're actually costing, with no AI sales pitch attached.
- Scored 9–14? Pick your single highest-scoring, narrowest use case and pilot it on a fixed 60–90 day timeline with a defined kill metric. Don't fund three initiatives at once. If the gap is specifically "we don't know which use case to pick or whether the data supports it," a formal AI readiness assessment turns this checklist into a scored, facilitated audit with a written gap list.
- Scored 15–20? You're past the readiness question. The work now is sequencing and execution. A full AI strategy engagement can rank your use-case portfolio by ROI and build the roadmap; if you already know what to build, go straight to an AI automation agency for operational workflows or an AI development company for custom systems, LLM applications, or ML models.
Frequently Asked Questions
What is an AI readiness assessment?
An AI readiness assessment is a structured evaluation of whether a business's data, processes, people, technology, and budget can support an AI initiative before money is spent building or buying one. It typically scores specific yes/no criteria across those five areas and converts the result into a go/no-go decision plus a prioritized list of gaps to close.
How do I know if my business is ready for AI?
Score yourself against a structured checklist, like the 20-point one above, across data, process, people, tech stack, and budget/governance. A score of 15 or higher across those areas signals real readiness; a score under 9 means the priority is fixing fundamentals, not evaluating AI vendors yet.
What's a good AI readiness score?
On a 20-point checklist like this one, 15–20 signals you're ready to scale AI investment with a sequenced roadmap. 9–14 supports one narrow, well-scoped pilot. Below 9, most of that budget would go toward fixing data and process gaps rather than toward AI that delivers value.
What's the biggest reason AI projects fail?
RAND's research into AI project failures points to organizational causes over technical ones: no shared definition of success, weak data foundations, and fading executive sponsorship after the pilot stage. MIT's 2025 research on generative AI pilots found the same pattern: 95% failed to show measurable P&L impact, mostly from failing to integrate AI into real workflows.
How long does an AI readiness assessment take?
Self-scoring this checklist takes about 20–30 minutes. A formal, facilitated readiness assessment, reviewing actual data, systems, and stakeholder interviews rather than self-reported answers, typically takes one to two weeks and produces a written scorecard and gap list.
Should a small or mid-size business even bother with a readiness assessment?
Yes, arguably more than a large enterprise, since a failed AI pilot costs a small business a much larger share of its budget and credibility. A 20-minute checklist costs nothing and routinely prevents a five- or six-figure mistake; skipping it is the expensive option, not the assessment itself.
Is a free checklist enough, or do I need a formal AI readiness assessment?
For an initial go/no-go decision, this checklist is enough. It becomes worth paying for a facilitated assessment once you're scoring in the pilot-ready range and need someone to audit your actual data and systems (not self-reported answers), rank multiple use cases by ROI, and build a sequenced roadmap you can execute or hand to a vendor.
