Skip to content
Free assessment

AI risk assessment for marketing teams

Eight questions on how your team actually uses AI. You get an exposure tier, the two or three places your risk is concentrated, and the approval workflow that fits — not a generic policy template. It scores in your browser and nothing is stored.

0 of 8 answered
–/24
Usage

How widely is AI already used across your marketing work?

Claims

Does your marketing make performance, pricing or comparative claims?

Regulation

How regulated or claim-sensitive is your sector?

Data

What kind of information gets pasted into AI tools?

Policy

Is there a written rule for what AI may and may not be used for?

Review

What happens before AI-assisted content is published externally?

Evidence

If a published claim were challenged, could you show what evidences it?

Brand voice

Since adopting AI, has your content held its distinctiveness?

The four tiers

Exposure sets the workflow, not the other way round

Governance that costs more than the risk it removes does not get followed, which is worse than none at all because it creates a false record of control. So the approval model scales with what is actually at stake.

Score 0–5

Contained exposure

AI use is limited, or the controls around it are already doing their job. The risk here is complacency rather than exposure: as usage widens, the informal checks that work today stop scaling. Keep the rules light, write them down, and revisit them when adoption steps up.

Workflow: Light-touch approval

Score 6–11

Moderate exposure

AI is embedded enough to matter and the controls have not caught up. Nothing here is alarming on its own, but the combination — daily use, external publishing and no documented boundary — is how unevidenced claims and quiet data leaks happen. A single reviewer and a short evidence habit closes most of it.

Workflow: Single-reviewer approval

Score 12–17

Elevated exposure

You have real exposure: AI in regular use, claims or regulated content in play, and no reliable record of what was checked. This is where a challenged claim becomes a scramble and where confidential detail has probably already been pasted into a public tool. Governance here is genuinely urgent, and cheap relative to the risk.

Workflow: Tiered approval with recorded evidence

Score 18–24

Critical exposure

Heavy AI use, claim-sensitive output, sensitive data in public tools and no gate anywhere. The realistic assumption is that something has already gone out that you could not defend if it were questioned. Fix the data boundary and the publish gate this week; the documentation can follow.

Workflow: Gated approval with an audit trail

Free starter pack

The AI Governance Starter Pack

The templates that turn your result into something operational: a one-page AI use policy, a risk register with starter rows already written, and a tiered approval workflow with the pre-publication checks to go with it.

  • A one-page AI use policy your team will actually read
  • A risk register with six starter rows most marketing teams need
  • A three-tier approval workflow sized to real exposure
  • A six-question pre-publication checklist for your content brief
  • A session plan for completing the pack with the people who publish

PDF · 10 pages · templates, not legal advice. Your details are used to send the pack and nothing else.

Questions

Questions about the AI risk assessment

What does the AI risk assessment actually measure?
Eight things that decide how exposed a marketing team is when it uses AI: how widely AI is already in use, whether your content makes performance or comparative claims, how regulated your sector is, what information gets pasted into AI tools, whether a written rule exists, what review happens before publishing, whether claims have recorded evidence, and whether your brand voice has held. Each answer scores nought to three, and the total places you in one of four exposure tiers.
How is the exposure tier calculated?
The eight answers add to a score out of twenty-four. Nought to five is contained exposure, six to eleven moderate, twelve to seventeen elevated and eighteen to twenty-four critical. The tier then sets the approval workflow: light-touch self-checking at the low end, a named reviewer in the middle, and two reviewers plus a retained evidence record where claims are regulated or commercially weighted.
Why does the recommended approval workflow have tiers rather than one rule?
Because uniform approval is the mistake that kills governance. Route everything through one reviewer and the team builds a workaround within a month. Tiering by exposure means internal drafting needs no approval at all, external content gets one reviewer, and only claim-heavy or regulated content carries a second review with recorded evidence. The gate holds where the risk is and disappears where it is not.
Is anything I enter stored or sent anywhere?
No. The quiz scores entirely in your browser and nothing is submitted — there is no email gate on the result. If you want the templates that go with it, the AI Governance Starter Pack is a separate download you can request with your email.
Is this legal advice?
No. It is a practical marketing governance assessment built from how AI actually gets used inside SME marketing teams. It will tell you where your exposure sits and what control closes it, and the output is written to be handed to a legal adviser or data protection officer for review where you have one.
Do you work with small businesses and sole traders?
Yes. MCG offers agency-style delivery with the flexibility of working directly with a freelance marketer. You do not need a marketing department or a large turnover to start. A website rebuild, a focused campaign or practical advice can be a standalone piece of work.
How much is a full website rebuild?
The standard website rebuild is £1,000 fixed for up to 40 pages, including structure, design, core page copy, technical SEO, enquiry forms and launch. Care is optional at £100 a month. Ecommerce, bespoke applications, complex integrations and third-party costs are quoted separately before work starts.
Can I hire you for a one-off project?
Yes. You can book a website rebuild, an SEO project, campaign work or a marketing review without taking an ongoing retainer. We agree the work and the price in writing first. Ongoing support is available if you need it afterwards.
Who will actually do the work?
You work directly with Samuel McGarrigle, founder of MCG Growth Marketing, with an MSc in Digital Marketing Management and MCIM membership. Samuel plans and delivers the work, and can work alongside your existing team or suppliers when needed.
Where do you work?
MCG is based in Cheshire and works with businesses across Staffordshire, Shropshire and the wider North West, with remote support available UK-wide.
Talk it through

Want the governance built rather than described?

The assessment tells you where the exposure is. A governance engagement puts the policy, the register, the brand-safety standards and the approval workflow in place with the team that has to use them.