AI Marketing Readiness: 12 Questions Before You Invest

How to use this

Answer each question honestly for your business as it is today, not as you plan it to be. Yes scores 2, partly scores 1, no scores 0. The maximum is 24. It takes about ten minutes, and it is most useful when two people answer separately and compare.

The questions come in the order a sensible investment would be built: first what you want, then whether you can measure it, then whether the work can be briefed, staffed and governed.

Assessment

AI marketing readiness: 12 questions

Answer for your business as it is today. Yes = 2, partly = 1, no = 0. Score 2 for yes, 1 for partly, 0 for no.

1. 1. Can you name the one commercial outcome you want marketing to move in the next 90 days?

   Without a single outcome, AI output cannot be judged and effort spreads thin.

   YesPartlyNo
2. 2. Do you know the current figure for that outcome?

   A baseline is the only way to tell later whether the investment worked.

   YesPartlyNo
3. 3. Are your key conversions tracked, and do the tracked numbers roughly match your own orders or CRM records?

   Automated optimisation towards a mis-tracked number makes results worse while reports look better.

   YesPartlyNo
4. 4. Is there a written description of your customer, offer and proof that someone new could work from?

   AI output is only as specific as the context it is given.

   YesPartlyNo
5. 5. Do you have documented brand voice guidance or a set of approved examples?

   Without it, AI drafts drift towards generic copy that could belong to anyone.

   YesPartlyNo
6. 6. Can you name the repeatable marketing task that takes the most time each week?

   Repeatable, frequent tasks are where AI returns real hours; vague pain is not a use case.

   YesPartlyNo
7. 7. Is a named person responsible for checking AI-assisted work before it goes out?

   Someone must own accuracy, claims and tone; tools do not carry accountability.

   YesPartlyNo
8. 8. Do you know which customer data you may use for marketing, and on what basis, such as consent or the soft opt-in?

   Automation that uses the wrong list breaks the rules at scale.

   YesPartlyNo
9. 9. Have you decided what AI must never do without approval, such as spending, publishing or sending?

   Limits set in advance are what allow more work to be delegated safely later.

   YesPartlyNo
10. 10. Can someone give 30 to 60 minutes each week to review results and decide the next change?

    Without a review rhythm, output accumulates and nobody connects it to results.

    YesPartlyNo
11. 11. Is there budget to run the work long enough to measure, typically at least one quarter?

    Stopping after a few weeks usually ends the work before any result can be read.

    YesPartlyNo
12. 12. Have you written down what result, three months from now, would count as success?

    A success criterion agreed in advance prevents moving the goalposts later.

    YesPartlyNo

Answered 0 of 12. Your result appears when every question is answered.

What each score range means

• 0 to 8: Not ready to invest yet. The foundations that make AI marketing measurable and safe are mostly missing. Spending now would likely buy activity rather than results. Next step: Spend the next month on questions 1 to 3 and 8: choose the outcome, record its baseline, fix tracking and confirm your data permissions.
• 9 to 16: Partly ready. You have some foundations but at least one gap would limit or distort results, usually context, capacity or governance. Next step: Fix any gating question scored 0, write the context document if question 4 or 5 scored low, then start with one workflow rather than several.
• 17 to 24: Ready to invest. Outcome, measurement, context and governance are in place. The main risk now is trying to do too much at once. Next step: Choose the single workflow tied to your outcome, agree the success criterion, and run it for a full quarter before extending.

What your band means

The band boundaries are our editorial judgement, not thresholds from a validated study. Use them to decide what to fix first, not as a benchmark against other businesses.

0 to 8: not ready to invest yet

This is a common and useful result. It usually means the business has not yet decided what marketing should achieve in numbers, or cannot measure it reliably. AI cannot fix either. The cheapest improvement available is often fixing tracking and writing down the outcome, and neither needs any AI spend.

9 to 16: partly ready

Most businesses land here. The typical pattern is a clear outcome and some measurement, but thin written context and nobody with time to check work each week. Investing now can work if you start narrow: one workflow, one outcome, one owner. Close the gaps in parallel.

17 to 24: ready to invest

You have the ingredients. The failure mode at this level is ambition: automating five things at once and being unable to tell which one moved the number. Pick the workflow closest to your outcome and prove it first. The guide to building an AI marketing system sets out that sequence.

The four gating questions

A high total can hide a zero that matters more than the rest. Treat these four as gates: if any scores 0, fix it before investing, whatever your band.

• • Question 1 (outcome). Without it nothing else can be judged.
• • Question 3 (tracking). Automated bidding and reporting optimise towards whatever is tracked. If that is wrong, the system gets better at the wrong thing.
• • Question 7 (a named checker). Someone must own what goes out.
• • Question 8 (permissions). The Information Commissioner's Office explains that marketing email to individuals needs their consent or all the conditions of the soft opt-in, which does not cover bought-in lists. Automation does not change that; it only changes how many people are affected.

Why these twelve

Gartner's July 2024 press release, headlined with the prediction that 30% of generative AI projects would be abandoned after proof of concept by the end of 2025, named poor data quality, inadequate risk controls, escalating costs and unclear business value as the reasons. The questions map onto those causes: 1, 2 and 12 address unclear value; 3 addresses data quality; 7, 8 and 9 address risk controls; 10 and 11 address whether the cost and effort can be sustained long enough to see a result. Questions 4, 5 and 6 cover the practical inputs that decide whether AI output is any good.

After scoring

If you scored in the lower two bands, the AI marketing audit turns the gaps into a list of specific fixes with evidence. If you scored well and would rather have the work run for you than build it, AI for Marketing provides a managed service, and a free written breakdown of your public marketing is the usual first step.

FAQ

Should I invest if I score in the middle band?

Possibly, if none of the four gating questions scored zero and you start with one narrow workflow. Treat the missing items as part of the first month's work rather than reasons to wait indefinitely.

Why does a named checker matter if the AI is good?

Because fluent output can still be wrong: an invented statistic, an outdated price, a claim you cannot support. Someone has to own accuracy, and that responsibility cannot be passed to software.

How often should I retake the assessment?

Once before investing and again after the first quarter. A rising score on questions 3, 4 and 10 is usually a better sign of progress than any single campaign result.

Does a high score mean AI will work for my business?

No. It means the conditions for measuring and controlling AI work are in place. Whether a specific workflow pays off is something only a measured trial can show.