By Dean McCoubrey Chief AI Strategist
|
AI marketing agency selection is one of the most consequential decisions a CMO will make in the next 12 months. And right now, it is also one of the easiest decisions to get wrong.
According to Salesforce’s State of Marketing 2026, 87% of marketers now use generative AI in at least one workflow. That number sounds impressive. It also means that claiming to “use AI” tells you almost nothing about an agency’s actual capability.
The gap between AI adoption and AI capability is where most agencies are quietly bluffing.
As AI systems increasingly mediate discovery, visibility will belong less to the loudest brands and more to the clearest ones. The same is true of agency selection. The question worth asking is not whether an agency uses AI. It is whether they have built something around it.
The reality behind the numbers:
- Writer’s 2026 enterprise AI report found that 59% of companies now invest over £1 million annually in AI, yet only 29% report significant returns.
- Supermetrics’ Marketing Data Report 2026 found that while adoption pressure is high, fewer than 6% of organisations have fully embedded AI across their operations.
- Jasper’s State of AI in Marketing 2026 found that only 41% of marketers can demonstrate AI ROI, down from 49% the previous year.
This is not a technology problem. It is a capability and judgement problem.
The seven questions below are designed to help CMOs and senior marketing leaders evaluate AI agency partners with the kind of commercial rigour the decision deserves. They are not a gotcha checklist. They are a framework for recognising genuine operational maturity.
1. Do they have a defined AI methodology or just AI tools?
In an AI-saturated market, access to tools is no longer the differentiator. Competitive advantage increasingly comes from how an agency combines AI capability with human judgement, strategic clarity, and operational discipline.
Any agency can show you a tool. Fewer can show you how they think.
Ask how AI is embedded across their process: strategy, research, production, quality review, and iteration. A credible partner should be able to walk you through a repeatable system, not just a list of platforms they subscribe to.
What to ask:
- How does AI inform your strategic recommendations, not just your content production?
- Where does human judgement sit within your AI workflow?
- How do you quality-check AI outputs before they reach a client?
- Can you show me a process document, not just a pitch deck?
What good looks like: A visible operating system with named stages, clear decision points, and human oversight built in. The agency should be able to explain what happens when AI gets something wrong, not just what happens when it goes right.
2. Can they show you proprietary IP, not just licensed software?
Licensed tools are accessible to every agency in the market. They create speed, but they rarely create differentiation. The stronger buying signal is whether an agency has translated its experience into proprietary thinking: frameworks, diagnostics, prompt systems, scoring models, or repeatable processes that are genuinely theirs.
Proprietary IP is what separates an agency that has done the thinking from one that has simply adopted the tooling.
| Licensed tools | Proprietary IP |
|---|---|
| Available to all agencies | Built from the agency’s own experience |
| Creates speed | Creates consistency and originality |
| Replicable by competitors | Defensible and compounding |
| Reflects vendor capability | Reflects agency judgement |
| Depreciates as tools evolve | Strengthens over time |
What to ask:
- Do you have proprietary frameworks, models, or diagnostics you can show us?
- How is your approach different from an agency using the same tools?
- What have you built that a competitor could not simply replicate by buying the same software?
What good looks like: Documented methodology that shapes how the work gets done, not just a slide deck listing third-party platforms. The agency’s IP should be visible in how they think about your brief, not just in what tools they mention.
3. How do they structure the team around AI capability?
AI capability should be visible in team design, not just in pitch language. The way an agency structures its people, roles, and workflows tells you whether AI is genuinely embedded or simply bolted on for appearances.
Strong teams use AI to amplify human capability. Strategy, accountability, and commercial judgement remain with people. AI expands what those people can execute, not what they are responsible for.
What to ask:
- Who owns strategy in your team, and how does AI inform their work?
- Who is responsible for quality oversight on AI-generated outputs?
- How are your team’s roles evolving as your AI capability develops?
- Where does the accountability sit when AI produces something that misses the brief?
This question also matters for your own planning. How an agency structures its team around AI gives you a useful reference point for how your own marketing function might evolve.
What good looks like: Clear separation between strategic ownership, AI-augmented execution, and quality oversight. Human judgement at the centre of decisions that carry commercial or reputational weight. AI fluency distributed across the team, not concentrated in one “AI person” who acts as a gatekeeper.
4. What does their measurement framework look like?
This is where the gap between AI theatre and AI capability becomes most visible. An agency that cannot connect its AI use to commercial outcomes is optimising for activity, not growth.
The important thing is whether the agency gets smarter as the engagement progresses, or simply faster.
What to ask:
- What metrics do you track before, during, and after AI deployment?
- How do you connect AI activity to commercial outcomes such as pipeline, conversion, or efficiency?
- How do you distinguish between outputs increasing and performance improving?
- What does a performance review with your team actually look like?
A basic scorecard to apply:
| Signal | Weak answer | Strong answer |
|---|---|---|
| Metrics | Output volume, content produced | Pipeline quality, conversion rate, CAC |
| Attribution | “AI helped us move faster” | Specific performance deltas with before/after data |
| Learning | No structured review process | Regular commercial review tied to strategy |
| Accountability | Reporting on activity | Reporting on outcomes |
What good looks like: A framework that tracks both performance and learning, so the agency becomes a more commercially intelligent partner over time, not just a more productive one.
5. How do they handle data, privacy, and responsible AI use?
Responsible AI adoption is no longer a legal footnote. It is part of commercial risk management, and increasingly part of how enterprise buyers assess agency maturity.
According to ThinkBRG, data and AI laws have grown by 400% since 2016. More than 20 US states now have AI-specific legislation, and EU AI Act enforcement is underway. Jones Walker’s guidance on responsible AI notes that clients are increasingly expecting agencies to explain data provenance, obtain explicit consent for AI training use, and provide clear documentation of how data moves through their systems. DLA Piper has noted that the FTC has taken a strong stance against misleading AI claims, making inflated capability statements a governance risk, not just a credibility one.
Red flags to watch for:
- Vague answers about which AI models are used and how client data is handled
- No documented policy on AI training exposure or data retention
- Inability to explain where client content or data goes after it enters an AI system
- No named person responsible for AI governance within the agency
Many agencies still treat governance as a legal inconvenience. Enterprise buyers increasingly see it as a signal of seriousness.
What good looks like: A documented framework covering data provenance, consent, storage, model usage policies, and accountability. Agencies that handle governance well are not just safer partners. They are more mature ones, and that maturity shows up in the quality of the work.
6. Can they demonstrate commercial outcomes, not just outputs?
AI is genuinely valuable when deployed with intelligence and commercial discipline. The question is whether your agency can prove it.
More content, faster production, and higher automation rates are not evidence of value unless they translate into growth economics. Writer’s 2026 enterprise AI report found that while 59% of companies invest over £1 million annually in AI, only 29% report significant returns. That gap lives in the space between outputs and outcomes.
What to ask:
- Can you show us case studies where AI use improved a specific commercial metric?
- Where has AI changed pipeline quality, conversion rate, or cost of acquisition for a client?
- How do you separate the performance impact of AI from other variables?
- What happened in an engagement where AI did not deliver the expected return?
A proof checklist:
- Named client outcomes tied to specific metrics
- Before and after data, not just directional claims
- An honest account of where AI added value and where it did not
- Evidence that the agency learns from performance, not just reports it
What good looks like: An agency that can point to specific engagements where AI improved growth quality, not just output volume. The strongest partners will also be able to tell you what did not work and what they changed as a result.
7. Do they have a distinctive point of view on AI, growth, and the future of marketing?
The strongest agencies in any era tend to have a clear, defensible perspective on where their field is heading. In an AI-saturated market, that matters more than ever.
A distinctive point of view is evidence of original thinking. It suggests the agency is doing genuine intellectual work, not simply repeating software narratives or repackaging vendor content. It also tells you something about the quality of strategic counsel you will receive.
What to look for:
- Published frameworks, not just trend commentary
- Evidence of original research, experiments, or lessons from client work
- A clear position on how AI changes growth systems, not just production workflows
- Thinking that helps clients navigate the future with more clarity, not more anxiety
Publishing frequency is not the signal. Depth of thinking is.
An agency that has genuinely grappled with questions like “what does AI mean for how brands build authority?” or “how does human judgement create competitive advantage in an AI-augmented market?” will show it in how they talk about your brief, not just in how many articles they have published.
What good looks like: A body of thinking that is specific, commercially grounded, and consistent over time. If their published work could have been written by anyone with access to the same AI tools, that tells you something. If it could only have been written by them, that tells you something better.
For a deeper look at how AI-first and AI-augmented agencies differ in practice, the AI-first vs AI-augmented guide is worth reading alongside this checklist.
A note on marketing IP: what it is, why it matters, and how agencies build it
Marketing IP appears in Question 2, but it deserves a slightly fuller treatment because it is one of the most misunderstood concepts in agency selection.
Marketing IP defined: The codified intellectual advantage an agency has built through repeated application, structured experimentation, and refined thinking. It includes frameworks, proprietary processes, diagnostic tools, scoring models, prompt systems, and any repeatable methodology that improves the quality and consistency of the work.
It matters for three reasons:
- Originality. As AI tools become more accessible, generic output becomes easier to produce and harder to differentiate. Agencies with proprietary thinking produce work that reflects genuine judgement, not just efficient processing.
- Consistency. IP-backed processes create more predictable outcomes. The agency is not reinventing its approach for every brief. It is applying a refined system that gets stronger with each engagement.
- Commercial intelligence. The strongest marketing IP connects creative and strategic decisions to commercial outcomes. It is not just a way of working. It is a way of thinking about growth.
For buyers, marketing IP is a useful proxy. If an agency can show you documented, proprietary methodology that shapes how they approach your brief, that is a strong signal that the thinking behind the work is genuinely theirs.
If they cannot, the tools probably are.
Frequently asked questions
How can a CMO tell if an agency is AI-washing?
Look for operational proof rather than pitch theatre. A credible agency can explain its methodology, show proprietary IP, describe its governance framework, and connect AI use to commercial outcomes such as pipeline quality, conversion rate, or cost efficiency. Vague answers to any of these questions are a signal worth taking seriously.
What is marketing IP in an agency context?
Marketing IP is the agency’s codified intellectual advantage: frameworks, models, diagnostics, prompt systems, and repeatable processes built from experience. It matters because it creates consistency, originality, and stronger commercial outcomes than generic tool use alone.
Should an AI marketing agency use proprietary systems or third-party tools?
Both. Third-party tools provide infrastructure and speed. Proprietary systems provide the thinking and differentiation. The strongest agencies combine licensed software with their own frameworks, so the work is more consistent, more commercially useful, and harder for competitors to replicate.
What should an agency’s AI measurement framework include?
A baseline, agreed commercial metrics, a clear method for linking AI activity to outcomes, and a regular performance review. The goal is to see whether the agency improves its commercial intelligence over time, not just its output volume.
What questions should legal or procurement ask about responsible AI use?
Ask about data provenance, consent controls, model usage policies, training exposure, storage, retention periods, and who owns AI governance within the agency. If the answers are unclear, that is a maturity and risk issue, not just a legal one.
How should a B2B marketing team be structured around AI capability?
The strongest structures separate strategic ownership, AI-augmented execution, and quality oversight. Human judgement should remain central to strategy and accountability. AI expands execution capacity rather than replacing responsibility. How an agency has solved this problem internally is often a useful model for how client teams might evolve.
The buying decision behind the buying decision
Choosing an AI marketing agency is not simply a procurement exercise. It is a decision about how your business wants to grow, compete, and operate in an AI-shaped market.
The agencies worth partnering with are not the ones making the loudest claims about AI. They are the ones that have combined AI fluency with human judgement, built proprietary thinking into how they work, taken governance seriously, and can show you where it has made a commercial difference.
That combination is rarer than the market noise suggests.
These seven questions will not guarantee the right choice. But they will make a weak answer visible, and in a market full of polished pitches, that is a significant advantage.
Read next: For a broader look at how AI-first and AI-augmented agencies differ in their operating model and strategic approach, the AI-first vs AI-augmented guide is the natural companion to this checklist.