Are You Actually Ready for AI? (A Simple Operational Test)

Everyone wants to talk about AI. Very few want to answer the question that should come first: is your business actually in a state where AI can do anything useful?

This is not a theoretical exercise. It is a practical one. AI needs specific conditions to function - clean data, clear processes, connected systems. If those conditions are not met, your AI investment will produce impressive demos and zero operational value.

Here are ten questions. Answer them honestly. They will tell you whether you are ready, nearly ready, or nowhere close.

The test

1. Can you produce a single, accurate list of your customers from your existing systems - without manual intervention?

This tests whether your core data is unified. If your CRM, ERP, and finance system each hold a different version of your customer list, and reconciling them requires someone to spend a day in Excel, your data is not ready for AI. AI needs a single source of truth, and getting there is a data and systems integration job. If you do not have one, everything it produces will be built on contradictions.

2. When a customer places an order, how many times is that data entered manually?

This tests whether your systems are connected. If the same information is typed into three different platforms by three different people, you have integration gaps. AI cannot fix those gaps. It will just inherit the inconsistencies that manual re-entry creates.

3. Can you trace a single transaction - from initial request to final delivery - across every system it touches?

This tests end-to-end data traceability. If the answer requires someone to manually stitch together records from three or four platforms, your data is fragmented. AI that analyses operational performance needs a connected data trail. If that trail has gaps, the analysis will have blind spots - and you will not know where they are.

4. Do your systems share a common identifier for the same customer, product, or order?

This tests integration architecture at the data level. If your CRM uses one customer ID, your ERP uses another, and your finance system uses a third, you have no reliable way to connect records across platforms. AI needs to match data across systems. Without shared identifiers, it is guessing - and it will guess wrong often enough to be dangerous.

5. How long does it take to produce your monthly board report?

This tests the accessibility and structure of your operational data. If the answer is "days," that means the data exists but is scattered, unstructured, and requires manual assembly. AI can generate reports in seconds - but only if the data feeding it is already connected and trustworthy.

6. Do you know which spreadsheets are critical to your operation - and who maintains them?

This tests your visibility into shadow systems. If there are spreadsheets running key processes that only one person understands, you have undocumented operational logic that sits outside your systems. AI cannot access what it cannot see.

7. When one of your systems goes down, do your other systems continue to function normally?

This tests integration architecture. If a CRM outage means your finance team cannot invoice, or an ERP issue means your warehouse stops receiving, your systems are coupled without proper orchestration. AI layers added on top of this will be equally fragile.

8. When was the last time someone audited your core data for duplicates, gaps, or conflicts?

This tests whether data quality is a practice or an afterthought. If the answer is "never" or "I do not know," then your data is degrading without anyone measuring how fast. AI trained or operating on silently deteriorating data will produce outputs that get worse over time - and nobody will notice until the damage is done.

9. Are your data fields standardised across systems - same formats, same categories, same naming conventions?

This tests data normalisation. If your CRM uses "UK" and your ERP uses "United Kingdom" and your finance system uses "GB," you have a normalisation problem that will make every AI output unreliable. These seem like small things. They are not.

10. If you connected an AI model to your operational data today, would you trust its output enough to act on it without checking manually?

This is the question that matters most. If the honest answer is no - if you would need someone to verify every output before acting on it - then your data is not trustworthy enough for AI to add value. AI that requires manual verification for every output is not saving time. It is adding a step.

How to read your results

8–10 ticks: You are genuinely ready for AI. Your foundations are solid. The question is where to apply intelligence for the highest impact - and that is a good question to be asking.

5–7 ticks: You are close, but there are gaps. The gaps are probably in data consistency and system integration. Fix those first through focused IT and process strategy - it is weeks of work, not months - and then AI becomes a practical conversation.

2–4 ticks: You are not ready. That is not a criticism. It is where most mid-market businesses sit. The right move is to invest in operational readiness - mapping processes, structuring data, connecting systems - before spending anything on AI.

0–1 ticks: You have bigger problems than AI. But that is also fine, because the work that fixes your operations will deliver value long before any AI is involved. Start with the foundations.

What to do with this information

If you scored lower than you expected, the instinct will be to feel behind. Resist that. Most businesses are in the same position. The ones that pull ahead will not be the ones that rush to adopt AI regardless. They will be the ones that do the foundational work first and get it right.

The path is clear: map your processes, structure your data, connect your systems, automate your workflows. Then apply AI enablement where it can make a measurable difference.

That is not a slow path. It is the only path that does not waste your money.