AI Readiness Assessment: A Practical Framework

Every AI strategy engagement we run starts with a readiness assessment, and every one of them uncovers at least one uncomfortable finding leadership wasn't expecting. That's not a criticism of the organisations we work with — it's simply what happens when you look honestly at data quality, infrastructure and team capability instead of assuming they're fine because a demo worked in a meeting.

Why Most Readiness Assessments Miss the Point

A lot of readiness assessments amount to a checklist survey sent to department heads, who understandably rate their own function's readiness generously. The result is a report that says everyone is basically ready, which is rarely true and isn't actionable even when it happens to be accurate. A useful assessment has to combine self-reported input with independent verification — actually looking at a sample of the data, actually testing the infrastructure, actually interviewing frontline staff rather than only their managers.

The Four Dimensions We Assess

We structure every assessment around four dimensions. Data readiness asks whether the data needed for a given use case actually exists, is accessible, and is clean enough to use — not in theory, but in a sampled, tested sense. Infrastructure readiness asks whether the technical environment can support the proposed system: integration points, security requirements, latency constraints. Team readiness asks whether the people who'll build, operate and use the system have the skills and capacity to do so, or whether that gap needs to be closed first. Governance readiness asks whether there's a clear owner, a decision-making process and a risk framework for AI initiatives — or whether every project will need to invent this from scratch.

Scoring Without Sugar-coating

We score each dimension on a simple scale, but the value isn't really in the number — it's in the specific findings underneath it. 'Data readiness: 2/5' is not useful on its own. 'Customer records exist in three systems with inconsistent formatting and no unique identifier linking them' is useful, because it tells you exactly what needs to happen before the AI initiative can succeed. We deliberately write findings this way — specific and slightly uncomfortable — because vague, diplomatic findings get filed away and never acted on.

Turning the Assessment Into a Plan

An assessment is only valuable if it changes what happens next. We close every readiness assessment with a short, prioritised remediation list — the handful of gaps that would block the highest-value use cases — rather than an exhaustive list of every imperfection found. Fixing everything is rarely realistic or necessary; fixing the two or three things blocking your best use case usually is.

If you're considering an AI investment and want an honest picture of where you actually stand before you commit budget, this is exactly what our AI Strategy engagements start with.