The AI Readiness Refrain: Why It Won’t Stop (and Shouldn’t)

Quick Takeaways

• 95% of AI pilots fail to deliver business value, not because of the technology, but because the strategic foundations aren’t there (MIT NANDA, 2025).
• Most AI readiness advice is built for large enterprises. If you’re an SMB, the standard checklists assume resources you don’t have.
• AI readiness is a three-legged stool: data, operations, and technology must work in concert. Remove one and the initiative falls. Start from the technology and you’re building the seat before the legs.
• Start from survival. What does your business need to achieve? That defines the outcomes, which define the use cases, which reveal what your data and operations actually need.
• Your first step takes 30 minutes, zero budget, and no vendor. Three questions, your leadership team, honest answers. That’s your starting point.

Here We Go Again

If you lead a small or medium-sized business (SMB) and the words “AI readiness assessment” make you sigh, you’re not alone. The refrain has been playing for two years now. Every newsletter, every conference, every vendor pitch. And through all of it, you’re still expected to make the right call with less budget, less time, and less room for error than the companies writing those articles.

Most of that content leaves you with a checklist and no genuine next step. This piece won’t do that. Not because it has all the answers, but because it starts from where you actually are: trying to make a smart decision before committing resources you can’t afford to waste.

Here’s why I care about this.

A while back, a long-established French SMB in professional services came to us looking for help with AI. They’d already spoken to their usual technology partner and received a proposal: 30,000 to 60,000 euros for a set of out-of-the-box AI solutions. No validated strategy behind it. No defined business outcomes. No return-on-investment (ROI) projection. Just tools and a price tag.

That budget, with a clear business case behind it, would have been perfectly reasonable. Without one, it was a gamble, and not a small one for a company of that size.

We took a different path. Instead of starting with the technology, we started with the business: what did the company need to achieve, where were the real opportunities, and what would each potential solution actually mean in terms of measurable outcomes? The whole engagement was lighter, faster, and far more focused than the original proposal. When the strategy was approved, it was handed to a technical team to work on the implementation specifics. The client left with clarity about what to prioritise, what each option was worth, and where to start.

That experience, and dozens like it over the past twelve years, is what this article draws on. The AI readiness refrain isn’t going away. But how you respond to it can change.

The Numbers Behind the Noise

The AI readiness refrain isn’t just noise. There’s real data behind it, and it paints a picture worth understanding, especially if you’re making decisions with limited budget and no room for a do-over.

Why 95% of AI pilots don’t deliver

In 2025, MIT’s Project NANDA (opens in a new tab) reviewed over 300 publicly disclosed AI initiatives, interviewed representatives from 52 organisations, and surveyed 153 senior leaders. The headline finding: 95% of AI pilots fail to deliver tangible business value. Not because the technology didn’t work, but because the strategic foundations underneath it weren’t there.

For a large enterprise, a failed pilot is a write-off on a spreadsheet. For an SMB, it’s different. It’s budget that won’t come back. It’s internal credibility spent as the team that championed the initiative now has to explain what happened. And it’s a board or a leadership team that’s less likely to approve the next proposal, even if the next one is the right one.

The cost of getting it wrong isn’t just financial. It’s organisational trust.

A readiness gap that’s wider than expected

Cisco’s 2025 AI Readiness Index (opens in a new tab) found that only 13% of companies are fully prepared to deploy AI across the six pillars that matter: strategy, infrastructure, data, governance, talent, and culture. That’s not 13% of small businesses. That’s 13% of all companies, including the ones with dedicated AI teams and multi-million-euro budgets.

For SMBs, the picture is even starker. A 2025 survey by Service Direct (opens in a new tab) found that 62% of small businesses cite a lack of understanding of AI’s benefits as a barrier to adoption. 55% cite cost. And the most common response to “why aren’t you using AI?” is simply: “It’s not relevant to our business.”

That last answer is interesting. Because in many cases, it’s not that AI isn’t relevant. It’s that nobody has yet explained how it’s relevant in terms the business actually cares about. That’s a readiness problem, not a technology problem.

Why Most AI Readiness Advice Misses the Point

If you’ve looked into AI readiness before, you’ve probably encountered the assessments. Cisco has one. Microsoft has one. Most major vendors offer some version of a self-service checklist or maturity questionnaire.

These tools aren’t bad. But they’re built for a specific kind of organisation, one with a dedicated AI team, a data engineering function, a governance structure, and the bandwidth to act on a lengthy report. If that’s your company, they’re useful.

If you’re an SMB, whether 30 people or 200, the experience is different. The checklist assumes resources you don’t have. The recommendations require roles that might not exist yet. And by question ten, you’re not clearer on what to do. You’re just more aware of everything you’re missing.

That’s not an AI readiness assessment. That’s a gap list.

The real blockers aren’t technical

I’ve worked with dozens of SMBs over the past twelve years, evaluating whether AI had genuine value for their business. The first conversation was never about the technology. It was about what the business was actually trying to achieve, and whether the foundations were in place to support it.

The blockers that surfaced were rarely technical. They were strategic and operational:

• No clear connection between data initiatives and commercial goals
• Operations that depended on undocumented knowledge and informal handoffs
• Leadership alignment on the idea of AI, but not on what problem it should solve
• A vendor pitch that sounded compelling but didn’t map to any defined business outcome

“AI can’t just sit on top of your stack like a novelty add-on. Without integration into ERP, CRM, supply chain, and finance systems, it becomes a point of failure.”

— Andrea Hill, Forbes (opens in a new tab)

She’s right. But for many SMBs, the issue isn’t integration with systems. It’s integration with how the business actually works. And that starts well before any technology decision.

The Three-Legged Stool: What AI Readiness Actually Requires

There’s a simple way to think about this. A stool needs a minimum of three legs to stand. An AI initiative needs the same: data, operations, and technology working in concert. Remove one and the whole thing falls.

This isn’t a framework I sat down to design. It emerged from years of conversations with business leaders who were trying to make sense of AI for their company. In almost every case, the pieces were already in their heads, they just didn’t have a shape for them yet. Once the model clicked, you could see the shift. They knew how to ask better questions and drive more strategic conversations. Nobody teaches this at school. If you’re lucky, a mentor shows you at work.

Data, operations, and technology in concert

Data flows through operations. Operations rely on data and generate more of it. They’re inseparable, but most companies don’t treat them that way.

AI is meant to leverage data to support operations, mostly through automation in its current form. But here’s the catch:

• No defined operations means nothing to automate.
• No data readiness means nothing to automate with.
• And automation alone doesn’t guarantee value. Speeding up a broken process just produces broken results faster.

Technology is the third leg — the enabler. But it’s never the starting point. Choosing a tool before understanding what your data and operations can support is building the seat before the legs. It might look impressive for a moment, but it won’t hold weight.

The tech-first trap with data

Most companies treat data as a technology problem first. Something to migrate, store, clean, or centralise. From an IT perspective, that makes sense. But from a business perspective, it turns data into an ever-expanding cost centre with an ever-expanding list of risks attached compliance, cybersecurity, retention, access control, and very little visible return.

When data is treated as a strategic resource instead, connected to the business outcomes you’re trying to achieve, it starts generating value rather than just consuming budget. That shift from “data as a tool” to “data as a resource” is one of the most underestimated steps in readiness. And it rarely happens without someone asking the right questions.

Start from survival, not from technology

Almost every SMB I’ve worked with is ultimately navigating the same thing: survival in a competitive landscape that keeps shifting. What survival means is different for each company, with growth for some, efficiency for others, market relevance for others still. But it’s always the starting point.

Survival defines the business outcomes that matter. Those outcomes define the use cases worth exploring. And those use cases reveal what data, operations, and technology actually need to deliver together.

The chain runs like this:

Survival → Business outcomes → Use cases → Data + Operations + Technology (in concert)

Start from the technology end, “we need an AI tool” , and you’re working backwards. You’ll end up with a solution looking for a problem, a stool built from the seat down.

Start from what your business needs to survive, and everything else follows with more clarity and less waste. That’s where building a strategy that connects all three becomes essential.

What I’ve Seen Work (and Fail)

Patterns don’t need a logo to be credible. After more than a decade of working on AI projects, first in machine learning, deep learning, and data strategy before the current wave, and more recently in the Large Language Models landscape, certain things have become predictable.

The pattern behind every stalled initiative

The conversation started with a tool. A vendor demo. A trending capability someone saw at a conference or in a newsletter. The energy was high, the expectations were real, and the budget was on the table.

But underneath:

• Data was scattered across systems that didn’t talk to each other, treated as a technical line item rather than a business resource
• Operations were inconsistent with processes living in people’s heads, not in documented workflows that could be automated or improved
• The people asking for AI and the people who understood the business reality were often in different conversations entirely
• Nobody had asked the foundational question: what business outcome is this meant to serve?

The result was predictable. The pilot launched, the results were underwhelming, and the internal narrative shifted from “AI will transform us” to “we tried AI and it didn’t work.” That second narrative is harder to undo than the first one was to build.

What the companies that moved forward had in common

They started from a business question. Not “what AI tool should we buy?” but “what do we need to achieve in the next 12 to 18 months to stay competitive?” That question forced clarity on outcomes, which forced clarity on use cases, which forced an honest look at whether the data and operations could support them.

They were honest about what their data could do today, not what it might do after a transformation programme they couldn’t afford or a migration that would take a year.

They treated readiness as an ongoing discipline, not a one-off event. The assessment didn’t end with a report. It informed every decision that followed: what to prioritise, what to delay, what to stop entirely.

And they understood that “not yet” was a valid outcome. Knowing you’re not ready, and knowing specifically why, is a strategic advantage. It protects budget. It protects credibility. It protects the team’s trust in future initiatives.

Started with a tool or vendor demo:

• Data scattered, treated as a technical line item
• Operations undocumented, inconsistent
• AI champion and business leadership in different conversations
• "We tried AI and it didn't work"

Started with a business question:

• Data assessed honestly for what it can do today
• Operations mapped enough to know what's automatable
• Alignment on outcomes before any technology decision
• "Not yet" treated as a valid, strategic outcome

“Knowing you’re not ready — and knowing why — is itself a strategic advantage.”

The role that doesn’t go away

One thing held true across every successful engagement: someone owned the full lifecycle. Not just the assessment, and not just the implementation. The whole arc, from discovery through strategy, through execution, through the ongoing reality of making it work and adapting when things changed.

AI doesn’t eliminate the need for that kind of strategic oversight. It increases it. The technology is evolving fast, but the judgment about when to act, where to focus, and whether a given initiative is worth pursuing, and that’s more important now than it’s ever been.

For SMBs, that role is rarely a full-time position. But it needs to exist. Whether it’s someone internal with the mandate and the perspective, or an independent external partner whose incentive is your clarity, not a licence fee.

For more perspectives like this, follow Realign on LinkedIn (opens in a new tab).

Your First Step (Not Your Tenth)

If this article does one thing, let it be this: give you something concrete to do before you speak to a single vendor.

A 30-minute exercise for your leadership team

Block 30 minutes. Gather whoever owns the business direction, the operations, and the data (even informally). No vendor. No tool. No budget needed. Just an honest conversation around three questions:

1. What does survival look like for us in the next 12 to 18 months? Not aspirational goals. Real outcomes. What does the business need to achieve to remain competitive, solvent, and relevant?
2. Where does our data actually live, and is it connected to those outcomes? Not “do we have data”, every company does. The question is whether it’s treated as a strategic resource tied to what you’re trying to achieve, or as a technical asset accumulating cost and compliance risk.
3. Which of our operations could genuinely benefit from automation, and which ones are too undefined to automate? If a process lives in someone’s head and changes depending on who runs it, automating it won’t help. It will just make the inconsistency faster.

Write down the answers. That document, however rough, is your readiness starting point. It’s more useful than any vendor questionnaire because it’s grounded in your reality, not a generic benchmark.

When to get outside help (and how to choose wisely)

If that conversation produced clear answers and a direction your team can act on, run with it. You may not need external help right now, and that’s a perfectly good outcome.

If it surfaced more questions than answers, or exposed misalignment that your team can’t resolve internally, that’s when an independent advisor can add value. Not a vendor. Not a technology partner with a solution already in mind. Someone whose job is to help you define what you actually need, so that when you do engage a vendor or start a project, the brief is right, the scope is honest, and the initiative has a real chance.

The key word is independent. Before committing budget to a platform or a licence, invest in clarity. It’s cheaper, faster, and it changes everything that follows. If you’re not sure where to begin, starting with discovery is often the right first move.

The Refrain Stays. Your Approach to It Changes.

The AI readiness conversation isn’t going to stop. The competitive pressure keeps building. The regulatory landscape, the European Union (EU) AI Act (opens in a new tab) included, keeps evolving. Boards and leadership teams will keep asking the question: “What’s our AI strategy?”

But readiness doesn’t have to mean paralysis. It doesn’t have to mean another assessment that tells you what you already suspected.

It means making your next move a deliberate one. Starting from what your business needs to survive, not from what someone else needs to sell. Getting the foundations right, data, operations, and technology working together, before committing resources you can’t afford to waste.

“Engaging with someone independent first, someone whose only agenda is your clarity, avoids wasting budget, credibility, and time. Or worse.”

If that 30-minute conversation with your team raised more questions than it answered, a discovery call might be a useful next step. No pitch. Just an honest conversation about where you are and whether there’s a way forward.

Book a 30-minute discovery call (opens in a new tab)

Frequently Asked Questions

What is an AI readiness assessment?

An AI readiness assessment evaluates whether your business has the strategic foundations, data, operations, and technology working in concert, to adopt AI effectively. For SMBs, it should start from business outcomes, not technology checklists.

How long does an AI readiness assessment take?

A meaningful starting assessment can take as little as 30 minutes with your leadership team, using three focused questions about survival, data, and operations. A deeper independent assessment typically takes days to weeks, not months.

Do I need an AI readiness assessment if my business is small?

Yes, arguably more so. SMBs have less room for error than enterprises. An honest assessment protects budget, credibility, and internal trust by ensuring any AI investment is grounded in real business needs.

Not sure where to start? A 30-minute discovery call can help you figure out what your business actually needs before you spend anything on AI. Book a free discovery call (opens in a new tab)

Disclosure

Realign is an independent advisory practice. No affiliate links, no sponsored content, no tracking pixels. The thinking, research, and writing in our content are our own. Generative AI is used as an editing and production aid only. All diagrams and graphics are original. Opinions and recommendations are based on direct experience.