The Honest Guide to AI Readiness

There's a statistic that should make every business leader uncomfortable.

Eighty percent of AI projects fail. Not "underperform." Not "take longer than expected." Fail. RAND Corporation published that figure, and the industry collectively shrugged and kept buying tools.

Here's another one. In 2025, S&P Global found that 42% of companies had abandoned at least one AI initiative entirely. Not paused. Abandoned. And perhaps the most telling number of all — over half of businesses surveyed couldn't even list the AI systems currently running inside their own organisation.

We're not in an AI adoption crisis. We're in an AI readiness crisis. The gap between buying AI and being ready for AI is where fortunes quietly go to die.

I keep having the same conversation. A business leader calls, excited about a tool they've seen demoed. They want to "implement AI." They've got budget. They've got enthusiasm. What they don't have is any honest assessment of whether their organisation can actually absorb what they're about to buy.

This is that honest assessment. The one the vendors won't give you, because they're selling the tools. The one the consultancies gloss over, because they're selling the implementation. I'm going to give you the entire framework — the same one that would cost thousands in a consulting engagement — for nothing. Because the organisations that succeed with AI aren't the ones with the biggest budgets. They're the ones that were honest with themselves first.

"Readiness" isn't what you've been told

Every major vendor has an "AI readiness assessment" now. Microsoft has one. Avanade has one. SurveyMonkey will let you build your own. They're all, without exception, funnels. You take the assessment, you score poorly on something, and would you look at that — they sell the exact product that fixes your weakness.

I'm not saying those tools are useless. Some are genuinely well-constructed. But they're measuring readiness against their own product, not against your actual capacity to change.

Real AI readiness isn't about having the latest tools. It isn't about a glossy strategy deck with "AI-first" on the cover. It's about five things that sound profoundly boring but determine absolutely everything.

The five dimensions of AI readiness

### 1. Data readiness

Not "do you have data." Every business has data. The question is whether your data is clean, accessible, documented, and governed.

Most companies' data is a disaster, and I say that with genuine empathy because it's nobody's fault. Data accumulates like sediment. Twenty years of different systems, different teams, different naming conventions. Customer records duplicated across three platforms. Spreadsheets that only one person understands, and that person left in 2019.

Here's what data readiness actually looks like: Can you, right now, pull a clean list of every customer interaction from the last twelve months, with no duplicates, in a format that another system could read? If the answer is no — and for most businesses it is — then you're not ready for AI. You're ready for a data cleanup project. Which is less exciting, but infinitely more valuable.

AI models aren't magic. They're pattern recognition engines. Point them at messy, incomplete, contradictory data and they'll find patterns in the mess. They'll confidently give you wrong answers derived from garbage inputs. The old "garbage in, garbage out" principle didn't retire when machine learning arrived. It got promoted.

### 2. Process clarity

You need to understand your processes before you can improve them. This sounds obvious. It isn't.

I've sat in rooms with senior leaders who couldn't draw their own customer onboarding workflow on a whiteboard. Not because they're incompetent — because the process evolved organically over years, with exceptions bolted onto exceptions, and now it lives in the collective memory of six people who all describe it differently.

AI amplifies whatever it's pointed at. Point it at a clear, well-understood process and it'll make that process faster, cheaper, or more consistent. Point it at chaos and it'll amplify the chaos. Faster chaos is not an improvement.

Before you think about AI, map your core processes. Not the idealised version in the policy document. The real one. The one with the workarounds and the "oh, but if it's a Friday we do it differently" exceptions. That map is worth more than any AI tool you could buy.

### 3. People and culture

This is where most AI implementations actually die. Not in the technology. In the humans.

AI literacy needs to exist across the organisation, not just in the tech team. Your finance team needs to understand what a large language model can and can't do. Your operations manager needs to know why the AI's recommendation should be questioned, not blindly followed. Your customer service team needs to feel like AI is augmenting their work, not threatening it.

Leadership buy-in matters, but not the kind that most companies have. "We should do AI" isn't buy-in. Buy-in is the CEO understanding that AI implementation means changing how people work, tolerating a temporary dip in productivity during transition, and investing in training that feels slow but prevents disasters.

The pattern I notice is this: organisations that succeed with AI have leaders who are genuinely curious rather than performatively enthusiastic. They ask questions. They admit what they don't understand. They protect their teams from the pressure to adopt everything at once.

Change management isn't a buzzword here. It's the difference between a tool that gets used and a tool that gets resented. And a resented tool, no matter how powerful, is shelf-ware within six months.

### 4. Technical infrastructure

Can your systems talk to each other?

This isn't about having the most advanced tech stack. It's about having a connected one. APIs, integrations, cloud readiness — the plumbing that lets data flow between systems without someone manually exporting a CSV and emailing it to another department.

Many businesses are running critical operations on software that has no API, no integration capability, and no upgrade path. That's not a moral failing. It's the reality of systems chosen ten years ago for perfectly good reasons. But it does mean that before you layer AI on top, you need to honestly assess whether your infrastructure can support it.

The question isn't "are we on the cloud?" The question is "if we wanted to connect our CRM to our inventory system to our customer service platform, could we do it without a six-month project?" If the answer is no, that's your starting point. Not AI.

### 5. Governance and ethics

Who's responsible when AI makes a bad decision?

If you can't answer that question in under ten seconds, you're not ready. This isn't theoretical. The EU AI Act is already reshaping compliance requirements for any business operating in or selling to Europe. Governance is no longer optional.

What is your escalation path when an AI system produces a biased output? How do you audit automated decisions? What data are you feeding into third-party AI tools, and have you read the terms about how that data gets used?

These aren't questions for later. These are questions for before you start. Because retrofitting governance onto an AI system that's already running is like trying to install brakes on a car that's already on the motorway.

Ten questions that actually matter

Here's your self-assessment. Be honest. Score yourself one to five on each, where one is "we haven't thought about this" and five is "we have this handled."

One. Can you produce a clean, deduplicated export of your core business data within 48 hours?

Two. Do you have documented data governance policies that people actually follow, not just ones that exist in a folder somewhere?

Three. Can three different people in your organisation draw the same version of your key business processes?

Four. Does your leadership team understand the difference between generative AI, predictive AI, and agentic AI — and which ones are relevant to your business?

Five. Have you invested in AI literacy training for non-technical staff?

Six. Can your core software systems exchange data through APIs without manual intervention?

Seven. Do you have a named person responsible for AI governance and ethics?

Eight. Can you list every AI tool currently being used across your organisation, including the ones individuals signed up for with their work email?

Nine. Do you have a clear, documented process for evaluating new AI tools before they are adopted?

Ten. If an AI system made a decision that harmed a customer, do you know exactly what would happen next?

If you scored below 30, you're not ready for a major AI implementation. You're ready for foundational work. And that's genuinely fine.

If you scored between 30 and 40, you have the foundations. Start small. Pick one well-understood process and one well-scoped AI tool. Learn from it before you scale.

If you scored above 40, you are in a strong position. But you probably already knew that.

The AI fatigue problem

I need to acknowledge something. Many of you reading this are exhausted.

Every week there's a new AI tool. Every conference has an AI keynote. Every vendor has pivoted to "AI-powered." Your inbox is full of AI whitepapers. Your LinkedIn feed is wall-to-wall AI thought leadership that all says the same thing in slightly different fonts.

AI fatigue is real, and it's a genuine business risk. Not because the technology isn't valuable — it is — but because fatigue leads to one of two equally dangerous responses: buying everything out of fear of missing out, or buying nothing out of sheer overwhelm.

The businesses I see struggling most aren't the ones that ignored AI. They're the ones that bought seven different AI tools across six departments with no coordination, no strategy, and no way to measure whether any of it is working. They have AI. What they don't have is any idea what it's doing for them.

The antidote to AI fatigue isn't more AI. It's clarity. What are you actually trying to improve? What does success look like in terms your CFO would accept? Which of your current problems could genuinely be solved by automation, and which ones are people problems wearing a process disguise?

You don't need to adopt everything. You need to adopt the right thing, at the right time, for the right reason. That requires slowing down enough to think. Which is, ironically, the one thing the AI hype cycle makes hardest.

What "ready" actually looks like

I want to be clear about something. Readiness isn't perfection.

You don't need flawless data to start. You need data that's good enough for the specific use case you're targeting, with a plan to improve it. You don't need every employee to be an AI expert. You need enough literacy that people can use the tools safely and question the outputs intelligently. You don't need a complete governance framework. You need the basics in place and a commitment to evolving them.

Minimum viable readiness looks like this: one clear business problem, one well-scoped dataset, one team that's trained and willing, one tool that's been properly evaluated, one person accountable for outcomes, and one honest plan for what happens if it doesn't work.

That's it. That's the starting point. Not a twelve-month transformation programme. Not a company-wide AI strategy. One focused, well-understood pilot that teaches you more in three months than any strategy deck ever could.

The organisations that get AI right almost never start with the most impressive use case. They start with the most boring one. The one where the data is cleanest, the process is simplest, and the risk is lowest. They learn how their organisation actually responds to AI — not how it theoretically should — and they build from there.

The uncomfortable truth

Most AI readiness content on the internet exists to sell you something. An assessment tool, a consulting engagement, a platform subscription. That's how the industry works, and I'm not pretending to be above it.

But here's what I believe. The businesses that will thrive in the next decade aren't the ones that spent the most on AI. They're the ones that were most honest about where they stood before they started spending. The ones that did the unglamorous work — cleaning data, mapping processes, training people, building governance — before they bought the shiny thing.

You don't need to be perfect. You need to be self-aware. You need someone in the room willing to say "we're not ready for this yet" when everyone else is caught up in the excitement. That person isn't a blocker. That person is the most valuable one at the table.

AI is going to reshape how businesses operate. That much is clear. But the reshaping won't be evenly distributed. It'll favour the prepared. And preparation, it turns out, looks nothing like what the vendors are selling.

It looks like honesty. It looks like foundations. It looks like doing the boring work first.

And if you've read this far, you already know where to start.