NOVIXX BLOG
What an AI readiness assessment actually examines
A good AI readiness assessment examines five things: your business processes, the systems and tools already in use, the quality of your data, your team's capacity and skills, and practical constraints like budget, timeline and compliance. It then scores the opportunities it finds and sequences them into a roadmap of quick wins and longer-term work.
WHAT GETS REVIEWED
What exactly gets reviewed in an AI readiness assessment?
Every assessment we've seen work well covers the same five areas, examined in roughly this order.
Processes
How work actually happens today, not how it's described in a handbook.
Systems & tools
What software is in use, and how reliably it shares information.
Data
How complete, structured and trustworthy your existing records are.
Team capability
Who would use, trust and maintain the result day to day.
Constraints
Budget, timeline, compliance and other limits that shape what's realistic.
PROCESS REVIEW
What processes get reviewed, and how?
Reviewing processes means looking at how work is actually carried out today — step by step, including the workarounds and exceptions that never made it into any manual — not how it's assumed to happen on paper.
This typically combines structured interviews with the people doing the work day to day, and a direct look at a handful of concrete, recent cases. The gap between the official process and the real one is exactly where AI often creates or destroys value.
An intake process might officially take three steps. In practice it takes seven, because two systems don't share data and someone manually reconciles the difference every time — the extra four steps are invisible until someone actually watches the work happen.
SYSTEM REVIEW
Why do systems and tools get reviewed before any AI is proposed?
The systems review exists to answer one question: can the tools you already use reliably share the information an AI solution would need, or would that information have to be entered by hand every time?
This covers CRM, ERP, communication tools, and how data actually moves between them — through an API, a manual export, or not at all.
The point isn't to recommend replacing tools that already work. It's to understand what's actually there, and where the biggest reliability gaps sit, before deciding what needs to change.
DATA REVIEW
How is data quality actually evaluated?
Data gets evaluated on three practical questions: is it complete enough to be useful, is it structured consistently enough to be processed automatically, and is it trustworthy enough that a decision based on it would hold up.
That means sampling actual records rather than trusting a description of what the data "should" contain — checking for duplication, missing fields, and how much relevant information sits only in free text.
It's also worth distinguishing messy-but-fixable from a genuine blocker: data that legally can't be used for a given purpose, or that simply doesn't exist yet, changes what's realistic far more than data that's just inconsistent. For a closer look at how to make that call yourself, see How to tell if your data is ready for AI.
TEAM REVIEW
Why does team capability affect AI readiness?
A technically sound AI solution still fails if the people expected to use it don't trust it, don't have time to learn it, or were never asked whether it solves a problem they actually have.
This part of the assessment looks at who would be the day-to-day user, whether they were involved in defining the problem in the first place, and whether there's internal capacity — even part-time — to own the tool after it launches.
Owning it doesn't mean building it. It means noticing when it stops working well, and having someone whose job it is to flag that.
CONSTRAINTS
What practical constraints get factored in?
Budget, timeline and compliance requirements decide what's realistic, not just what's theoretically possible — the same opportunity can be the right or wrong next step depending entirely on what a business can actually commit to right now.
A healthcare or financial-services SME, for example, faces compliance requirements that materially change which solutions are viable and how data can be handled.
A business with no capacity to manage a multi-month project needs a different starting point than one that can dedicate a person part-time for a quarter — and a credible assessment says so explicitly, rather than recommending the same roadmap regardless.
HOW OPPORTUNITIES GET SCORED
How do opportunities get scored?
Each opportunity found during the review gets scored on three dimensions — business value, implementation feasibility and organizational readiness — and the combination determines where it lands in the roadmap. The AI Retrofit Check's methodology follows exactly this scoring approach.
Business value
Time saved, cost reduced, errors avoided, or revenue affected. The clearer and more measurable the impact, the higher the score.
Feasibility
How much technical and organizational effort the change actually requires, given the systems and data reviewed earlier.
Readiness
Whether the data, systems and team involved are mature enough today, or whether groundwork has to happen first.
A QUICK WAY TO SELF-ASSESS
How can you tell, roughly, where your own opportunity would score?
You don't need a formal assessment to get a rough read. A few honest answers about a specific process usually point in the right direction.
- "This is a metric we already track." — points to higher feasibility.
- "Only one person understands how this process works." — points to lower readiness.
- "It affects revenue or a compliance deadline." — points to higher business value.
- "We'd need to build a new system first." — points to lower feasibility, longer-term.
FROM SCORES TO A ROADMAP
How does a roadmap get sequenced from those scores?
Opportunities with high value and high feasibility become quick wins and go first. Opportunities with high value but low readiness become foundation work that has to happen before anything bigger. Everything else gets sequenced in between, by dependency.
- Quick win — high value, high feasibility. Scoped and started within weeks.
- Foundation work — high value, low readiness. Needed before larger initiatives can start.
- Long-term initiative — high value, low feasibility. Sequenced later, once dependencies are resolved.
- Not prioritized — low value regardless of feasibility. Parked, revisited only if circumstances change.
THE OUTPUT
What does a business actually receive at the end of an assessment?
At minimum, a completed assessment should produce three things: a clear picture of where the business currently stands, a ranked list of opportunities, and a sequenced set of next steps — in plain business language, not a slide deck full of jargon.
Specifically: a readiness score across the areas reviewed, an opportunity list or heatmap showing value against feasibility, and a roadmap that separates quick wins from longer-term initiatives. A credible assessment also names the gaps that aren't ready yet — not just the opportunities that are. To see this structure applied to an illustrative business, browse our sample report.
For readers who want a broader, vendor-neutral reference beyond a single assessment, NIST's AI Risk Management Framework is a free framework organizations use to structure AI governance more generally. It's a useful complement to a business-specific readiness assessment, not a replacement for one.
FAQ
Frequently asked questions about AI readiness assessments
How long does an AI readiness assessment take?
It depends on the size and complexity of the business, but a structured assessment for a typical SME is usually measured in days of effort spread over a few weeks, not months — most of the time goes into interviews and reviewing actual systems and data, not producing the report itself.
Do we need prior AI experience to benefit from one?
No — if anything, the assessment is most useful for businesses that haven't started yet, since it's designed to surface exactly the gaps that would otherwise only show up after a project is already underway.
What's the difference between a readiness assessment and just buying an AI tool?
Buying a tool starts with a solution and hopes it fits. A readiness assessment starts with your processes, systems, data and goals, and only then identifies which tools or approaches actually fit — which is why it typically produces several ranked opportunities rather than a single purchase decision.
Can a business run this kind of assessment internally?
Parts of it, yes — mapping processes and systems is something an internal team can do. The parts that are harder to do internally are usually the ones that benefit from an outside view: spotting gaps that feel normal because "that's just how we've always done it," and comparing opportunities against a consistent scoring method rather than internal politics.
Want this assessment applied to your own business?
The AI Retrofit Check works through processes, systems, data, team and constraints, scores what it finds, and hands you a practical roadmap — the same structure described in this article, applied to your business specifically.
Request AI Retrofit Check