AI Readiness Checklist for Marketing: 10 Questions

In late 2024 the OECD surveyed over 5,000 small and medium-sized businesses in Austria, Canada, Germany, Ireland, Japan, Korea and the UK. About 31% were using generative AI. Among those that weren't, 57% said it didn't suit their work, 52% worried about what happens to the information fed into the models, and 50% said their staff lacked the skills (OECD, 2025).

Those are the right worries. This checklist turns them into ten questions about your marketing that you can answer in one sitting.

What does AI readiness mean for marketing?

It means AI has something to speed up. It works best on a process that already exists: a repeatable task, a clear goal, data it can use, and a person who checks what it produces. Point it at a scattered process and you get the same mess, faster and with more confidence. Readiness is having those four things in place before you buy anything.

AI readiness checklist: 10 questions

Answer yes or no. Be strict: "sort of" counts as no.

Can you name the one number AI should move? Hours spent per campaign, qualified leads per month, days from brief to first draft. One number, written down.
Could you write down how the task is done today, step by step? If nobody can, the task isn't repeatable yet, and AI can't repeat it for you.
Is the data it will use in one place, with a source for every record? A customer list spread across three tools and a spreadsheet gives three versions of the truth.
Is there one named person who checks the output before a customer sees it? A committee doesn't count. Neither does "everyone".
Do you know what must never go into an AI tool? Client contracts, personal data, anything under NDA. Read the data policy of each tool before the first paste, not after.
Do you have three examples of your own best work to show it? A model copies what it's given. Without examples of good, it gives you average.
Can you measure the "before"? If you don't know how long the task takes today, you won't know whether AI saved anything.
Could you describe your positioning in two sentences? Who you are for, and why you rather than the alternative. AI can't write on-brand copy for a brand that hasn't decided what it stands for.
Who will look after it in month three? Prompts drift, tools change, a new team member doesn't know the rules. Someone has to own the upkeep.
What will you stop doing if it works? Saved hours that go nowhere aren't savings. Decide in advance where they go.

How do you score it?

Count the yeses. This is our rule of thumb, not a standard:

8 to 10: you are ready for a pilot. Pick one task and start.
5 to 7: fix the gaps first. They are usually questions 3 and 4, the data and the owner.
4 or fewer: not yet. The groundwork above will improve your marketing whether or not you use AI.

What should your first AI marketing pilot be?

One task, one owner, one number, four weeks. Good first pilots are tasks you already do often, that follow a pattern, and where a mistake is cheap to catch: first drafts of product descriptions, turning a long article into shorter posts, summarising a month of campaign reports. Poor first pilots are anything customer-facing and unsupervised, such as a chatbot answering refund questions on day one.

If a prompt keeps giving you generic output, our free Prompt Autopsy shows where it leaves too much to the model's imagination.

Questions

What is an AI readiness assessment?

A structured check, done before adopting AI, of your goals, processes, data, people and risks. Large companies run long versions with consultants. For marketing in a small business, the ten questions above cover the parts that decide whether a first pilot works.

Does a small business need an AI strategy?

It needs a first use with a clear number and a named owner. A written strategy is more useful after the first pilot, once you know what works in your business and what doesn't.

What are the risks of using AI in marketing?

Three come up most. Information leaving your control, which 52% of the SMEs in the OECD survey that hadn't adopted it were worried about. Output that sounds like everyone else's. And mistakes delivered with confidence. A named reviewer and a short list of what never goes in cover most of it.

Related