7 Signs Your Business Is Not Ready for AI Yet (And What to Do First)

    AI readiness isn't a personal failing or a sign your business is behind. It's a sequencing problem. The same MIT NANDA work that produced the 95% headline figure on failed pilots traces almost every failure back to the same handful of upstream gaps. Not the technology, the foundations underneath it. Process, data, people, measurement. When one or more of those is weak, the cleanest tool in the world drops into the gap and disappears.

    What follows are the seven signs we see again and again across mid-market UK businesses. None of them are about whether you're smart, ambitious, or technical enough. They're about whether the ground is firm enough for the build to land on. If three or more of these apply, stop the procurement conversation and fix the upstream gap first. The order at the end of this article tells you which one to start with.

    The seven signs

    1. You can't describe the problem in one sentence

    Every failed AI project we've inherited started with a brief that couldn't pass the one-sentence test. The conversation begins as "we want to use AI in marketing" or "we should be doing more with AI", and by the time procurement gets involved the original pain has gone missing. The result is a pilot scoped to fit a tool, not a workflow. A lot of the readiness work happens before any tool gets bought, and it starts here. If the leadership team can't write the brief without a vendor name in it, the brief isn't ready. A workable use case sounds like "score inbound leads inside 24 hours so the sales team stops chasing cold ones", not "AI-enabled marketing". The first one survives contact with reality. The second gets quietly redefined three times during the pilot.

    Next move: run the candidate problem through the AI Necessity Test. It forces the one-sentence brief in eight minutes, ahead of any procurement conversation.

    2. Your process isn't documented

    If nobody on the team can draw the current process on a single page, you don't understand it well enough to improve it with AI. We see this constantly. A marketing operation running on twelve tools, three spreadsheets, and a Slack channel of tribal knowledge, with no map of how a lead actually moves from form fill to opportunity. AI dropped into an undocumented process is just a faster way to produce inconsistent output. The pattern is universal across mid-market UK businesses. The work has grown organically, the team has worked around the gaps, and the actual flow lives in the heads of two or three people who joined three years ago. That's not an AI problem, it's a process visibility problem.

    Next move: a digital Gemba walk on one workflow. Screen-share with the team that runs it, write the steps down, count the handoffs. We cover the technique in the AI Readiness pillar. Two days of mapping unlocks more value than a quarter of tool evaluation.

    3. Your data lives in five different systems and nobody owns it

    The data readiness gap shows up in three layers and a lot of businesses are missing the bottom one. The data is accessible, the team can get to it without a six-week IT ticket. The data is clean, the records are consistent enough to act on. The data is governed, somebody owns it and the rules for using it are written down. Most mid-market businesses are decent on accessible, patchy on clean, and missing entirely on governed. The CRM has three sales cycles' worth of inconsistent stage names. The marketing platform has half-filled required fields and a duplication rate above 15%. Gartner has reported for years that poor data quality is the leading cause of AI project failure, and the mid-market version is smaller in absolute pounds and larger as a share of marketing budget.

    Next move: name one person whose job description includes "the data stays clean". Without that role, the cleanup is a one-off and the records drift again inside a quarter. The AI Readiness Audit produces a written ownership map as part of the deliverable.

    4. Nobody on the team has AI skills (or interest)

    Tools without trained operators is shelfware. The Deloitte AI Institute reported in 2026 that insufficient worker skills is the single biggest barrier to capturing value from AI tools, and we see the same pattern in every readiness audit we run. A marketing team of six, one person who's curious about AI in their own time, four people who've used ChatGPT once or twice, and one person who's actively suspicious of it. That mix won't run a pilot, never mind embed AI into the daily workflow. The fix isn't more demos or a vendor-led training day. It's getting the team's hands dirty on the specific workflow the tool will run, ahead of the launch, so the first four weeks don't produce a productivity dip that kills confidence.

    Next move: pick one person on the team and give them four hours a week for a month to learn the tool that will run the pilot. Not a course, the actual tool, on the actual workflow. We call that role the AI Champion and we cover it inside the readiness framework.

    5. Leadership thinks AI is an IT project

    The clearest sign of an AI project that won't ship is a leadership team that hands the file to IT and asks for a status update next quarter. AI sits across process, data, people, and outcomes, all four of which live with the operating leaders, not the technology function. A lot of the time the deflection is comfort-driven. The MD or Head of Marketing doesn't feel confident on the topic, IT does the procurement, and the project becomes a tool rollout with no operating sponsor. The Adecco Group reported in 2025 that 53% of CEOs struggle to align their executive teams on AI strategy, which is another way of saying that more than half the AI conversations inside mid-market businesses are happening one rung below where they need to.

    Next move: name the operating sponsor before the procurement starts. MD, Head of Marketing, or whoever owns the metric the project is meant to move. The right test is whether that person can describe the use case, the metric, and the kill criterion in one minute. If they can't, the project isn't ready.

    6. You don't have a baseline metric to improve

    No baseline, no ROI. A lot of pilots we walk into have produced something, a chatbot, a scoring model, an automated brief generator, and nobody can tell us what the workflow looked like before. The team didn't measure cycle time, or cost per lead, or hours spent at each stage, before the tool went in. So when the finance director asks for the return number at week 12, the answer is a story about "efficiency gains" with no figures attached. McKinsey's State of AI 2025 work captured this gap precisely: 88% of businesses have AI live in at least one function, only 39% report a measurable EBIT effect. The gap between adoption and impact is, in our experience, mostly a measurement-framework gap.

    Next move: pick one metric, measure the workflow as it runs today for two weeks, then start the pilot. Cycle time, cost per outcome, error rate, throughput, pick one. Two weeks of disciplined measurement on the front end saves the project at the back end.

    7. You've already tried once and it didn't work

    If the business has already tried AI once and the project quietly died, the next attempt has a higher bar to clear. The board has weaker patience, the team has the scars, and the budget owner has internalised "AI doesn't work here" as a defensible position. The honest read on a failed first attempt is that one of signs one through six was the actual cause, not the technology. We walk into businesses with this scar tissue regularly. The first move is a post-mortem on the dead project, not a fresh pilot. What was the brief, what was the metric, who owned it, what broke. Until that conversation happens, the next attempt is just the previous one with a different vendor name on the invoice.

    Next move: write a one-page post-mortem on the previous attempt before you scope the next one. The four-failure-pattern framework is a useful structure for that exercise. If the post-mortem doesn't reach a named root cause, the next pilot will hit the same wall.

    What to do first

    Reading seven signs and nodding at five of them is a familiar feeling and a bad starting point. Don't try to fix all seven at once. The sequencing matters as much as the work, and the wrong order will cost you a quarter. The seven signs map onto the four pillars we use in the AI Readiness pillar: Process, Data, People, Strategy. Signs one and two are Process gaps. Sign three is Data. Signs four and five are People. Signs six and seven are Strategy and measurement. Process always comes first, because a workflow nobody has mapped is a workflow nobody can improve.

    A 30-60-90 sequence that works at mid-market scale looks like this. Days one to thirty: fix the most upstream sign. If sign one or two applies, that's where you go. Write the one-sentence problem statement using the Three-Question Filter (what's the workflow, what breaks, what does fixing it look like in pounds). Map the current process with a digital Gemba on one team. No tool decisions yet, no vendor calls. Days thirty-one to sixty: the next pillar in your sequence. Usually Data, sometimes People. Name the data owner, audit the cleanliness, build the ownership map. Or pick the AI Champion and put four hours a week of their time on learning the tool that will run the eventual pilot.

    Days sixty-one to ninety: baseline the metric and scope the pilot. Pick one workflow, one metric, one decision rule. Cycle time on inbound lead qualification, 4.2 days today, target under 2 days at week 12, decision to scale or kill at week 14. That sentence is the difference between a pilot you can defend at the board meeting and a "learning experience" you can't. The work to produce it is unglamorous and uncomfortable, and a lot of the businesses we work with skip it and pay for the skip later. Pick the most upstream sign that applies and start there. The downstream ones get easier once the foundation under them is firm.

    Where to start

    Three options, in commitment order. The first is free and takes eight minutes. The second is a 14-day engagement that produces a written readiness map and a sequenced fix plan. The third is the full framework on this site. Pick whichever matches the depth you actually need.

    References

    • MIT NANDA. "The GenAI Divide: State of AI in Business 2025." MIT Media Lab Project NANDA, 2025. Source of the 95% no-measurable-return figure underpinning the readiness frame.
    • Gartner. "Data Quality and AI Project Failure." Ongoing research, 2024-2025. Source of the data-quality-as-leading-cause finding referenced in sign 3.
    • McKinsey & Company (QuantumBlack). "The State of AI." 2025. Source of the 88% adoption vs 39% measurable EBIT impact gap referenced in sign 6.
    • Deloitte AI Institute. "State of Generative AI in the Enterprise." 2026. Source of the worker-skills barrier finding referenced in sign 4.
    • Adecco Group. "Global Workforce of the Future: CEO Outlook on AI." 2025. Source of the 53% leadership-alignment finding referenced in sign 5.