What a good AI readiness audit includes
Here is what a useful AI readiness audit covers and why many recommendations are never acted on. We have also included a checklist to help you judge any audit before you buy it.
Clinton Onyekwere5 min read
Here's what we usually see. Many businesses buy AI tools first and ask where they help later. A proper audit reverses this. It shows you exactly where AI works, and where it doesn't, before you spend any money.
But not every audit does this. Some just list tools or score your general readiness, and many end in recommendations nobody acts on. This article sets out exactly what a useful audit covers.
Start with the real work
The thing is, people can only automate what they can describe. We find teams naturally forget the small steps they take every day. By watching recorded walk-throughs of recent tasks, we spot the hidden details that actually make a process work. It gives a much clearer picture than just asking questions.
McKinsey's 2025 State of AI survey found redesigning workflows had the biggest effect on earnings from generative AI. The thing is, only 21% of respondents using it have fundamentally redesigned theirs. This is why our audit maps your workflows first.
Audit people, processes and technology
A 2025 study by Heimberger and colleagues looked at 1,334 manufacturing firms in Germany. Only 28% reached high organisational readiness for AI, and just 17% were ready across both technology and organisation. The thing is, people and processes are the weaker side. An audit checking only systems and data misses it.
Rate every step. Know where AI stops
A good audit looks at your work and makes one of four choices for every step. You can buy a ready-made tool, improve your process with AI, build something new, or keep the work with your people.
The UK Government's AI Playbook reminds us to pick the right tool for the job. In our experience, you have to accept that AI is not always the best solution. Keeping a task with your people is a successful audit result.
Clear decisions someone owns
In a 2022 field scan by Costanza-Chock and colleagues, 152 algorithm auditors responded. It's telling that 80 per cent said their recommended changes to AI systems were never implemented.
So a few ranked choices beat a long list. From what we've seen, each recommendation needs a clear first move, a success measure, a known cost, and an owner.
Keep it proportionate
In 2026, the UK Department for Science, Innovation and Technology decided not to publish its AI Management Essentials tool, recognising concerns about its complexity for smaller organisations. A small business needs a light, focused audit that takes weeks, explains how data is handled, and leaves a report anyone can act on.
Questions to ask before buying
- 01Does it review actual recent work alongside the staff interviews?
- 02Does it check people and processes alongside systems and data?
- 03Does it rate each step and say what stays with people?
- 04Does each recommendation include a way to measure if it worked?
- 05Does it outline the costs and give a clear first step?
- 06Does it explain clearly how your data is handled?
- 07Can you use the report with any team or supplier?
In one to two weeks, we walk through your real work and rate every step. You get three ranked recommendations with success measures, a costed first move, data handling notes, and a report to keep. We provide a fixed quote after a short call.
Sources
Get our newest articles
We only email you when a new piece goes live.