What a GenAI readiness audit should include (and what it shouldn't)

Most companies don’t need more AI enthusiasm. They need a short, honest piece of work that tells them where Generative AI would actually pay for itself — and where it wouldn’t. That’s what a readiness audit is for.

Here is what a useful one includes, based on how we run them at Rawyll Group and inside our own e-commerce brands.

1. A map of your real workflows

Not your org chart — your workflows. Where do inquiries arrive? Who answers them, with what tools, in how many steps? Where does text get written, copied, re-typed? An audit that starts with models instead of workflows will recommend tools you don’t need.

The output of this step is boring on purpose: a list of the repetitive, text-heavy, rule-bounded tasks in each department. Those are the only places GenAI reliably earns its keep.

2. Use-case scoring against return, not excitement

Each candidate use case gets scored on three questions:

• Volume — does this task happen often enough to matter?
• Cost of error — if the AI gets it wrong, who is harmed, and would a human review step catch it?
• Line of sight to the P&L — can you name the metric this would move: response time, hours recovered, conversion, refund rate?

If a use case can’t answer the third question, it isn’t ready. Park it.

3. A data reality check

GenAI systems are only as useful as the context you can hand them. The audit should verify, concretely: can your order history, customer records, product data, or internal documents actually be retrieved by a system at the moment it needs them? “It’s in the ERP somewhere” is a finding, not a footnote.

4. A plain statement of where AI does not belong

Any audit that finds AI fits everywhere is a sales document. Expect — and demand — a section that says: these tasks stay human. Judgment calls, exception handling, anything where the error cost is high and review is impractical.

5. A ranked list, not a vision deck

The deliverable is one page that says: do this first, this second, skip these. Each item names the workflow, the metric it should move, and what a pilot would look like. If the deliverable is forty slides about the future of work, you paid for a keynote.

What happens after the audit

A good audit flows directly into a small pilot: the top-ranked use case becomes a working prototype, measured against one KPI you chose. That’s the second step of our four-step method — audit, pilot, integration, enablement — and it’s deliberately scoped so you can stop after any step and keep everything you learned.