This blog represents anonymised insights from a roundtable discussion. All participant information has been removed to protect individual privacy while preserving valuable learnings for the broader community. Prefer a PDF? Download the insights here.Â
Theme 1: Most organisations are stuck in the messy middle
CMOs across sectors are describing their AI position in almost identical terms: active experimentation, fragmented use cases, pressure from above, and no clear framework for deciding what to prioritise.
Formal tool policies are rarely the full picture. It is widely acknowledged that teams are using unapproved tools independently, creating ungoverned workflows that sit outside any security or data framework.
The structural tension is consistent. Boards push for adoption based on what they have seen through their own use of point solutions (like ChatGPT or Claude). IT and security teams carry the risk and are already stretched. Marketing leaders are caught in the middle, without a shared language for what AI strategy actually means inside the organisation.
Pressure without a plan creates misalignment
When enthusiasm from the top meets uncertainty in the middle, the result is activity without direction. Teams adopt and demonstrate AI use, but without defined outcomes or governance, it is almost impossible to know whether genuine value is being created.
The organisations that break out of this pattern are those that replace ad hoc experimentation with a deliberate, outcome-led approach.
The Artificial Intelligence Index Report:
“Leadership are experimenting with AI, but don’t have a clear roadmap or strategy to measure the impact and prioritise where it should actually drive ROI.”
Theme 2: Strategy starts with problems, not tools
The most common mistake in AI adoption is leading with the technology. The strategic approach is to start with a clearly defined business problem and ask whether AI is the right solution.
This reframe changes the criteria for success: instead of asking whether a tool works, the question shifts from whether a tool works to whether a defined problem has been solved and the impact can be measured.
The TRIP framework: time, repetition, importance, pain
A structured audit of where AI can create immediate value is more useful than a top-down strategy exercise. Evaluating potential projects against four filters, what is consuming time, what is repetitive, what is important, and what is causing operational pain, quickly identifies high-value opportunities and separates them from peripheral ones.
Innovation and AI specialist:
“When you get those four pieces in play, you can start to understand where the value of the next AI project works.”
Keep the first project small and scoped
There is a strong case for keeping early AI projects within a tightly defined budget, in the range of £6,000 to £30,000. A small, well-measured project produces something a large one rarely can: proof. Proof that the organisation can deliver, proof that it can measure the outcome, and proof that the investment was worth making. Without these in place, it is almost impossible to know whether AI has created genuine value.
CMO, attendee:
“I really like the idea of doing ring-fenced projects that prove out the case for AI in specific use cases. Otherwise, you never really know the true benefit and what you should scale up.”
Theme 3: The mindset that separates leaders from the rest
Tools and frameworks only go so far. The other part of the challenge is how leaders think about AI. The GAIN framework describes the mindset principles that distinguish organisations building durable AI capability from those that remain permanently in the messy middle.
Goal orientation: ask what matters most, not what AI can do
The shift from asking what AI can do to asking what matters most to the business right now changes everything about how projects are selected and measured. It keeps the focus on outcomes, not technology, and it is a far more credible position to hold in front of a board.
Abundance thinking: frame AI as a source of value
Fear of AI is real inside organisations, particularly among teams who see their own skills being replicated. Leaders who demonstrate AI’s value at an operational level, rather than positioning it as a replacement for individual capability, are far more likely to build willing adoption. AI’s analytical applications are often less threatening and more commercially powerful than its generative ones.
Iteration and nexus thinking: build for adaptability and wider impact
AI evolves faster than any five-year strategy can accommodate. Designing projects with iteration built in, and accepting that 70% is sometimes good enough, is not a compromise on rigour. It is a more realistic approach to a landscape that will have moved on before perfection is reached.
Equally, every AI implementation touches more than the immediate problem it solves. Tracing those connections through the wider organisation, including people, security and governance, is what separates considered leadership from reactive adoption.
Innovation and AI specialist:
“AI has scrambled the logic of traditional leadership. It moves too fast, evolves in unpredictable ways, and refuses to stay within the lines of a five-year strategic deck. Effective leaders make decisions in tight feedback circles and adjust to what is happening in real time.”
The bottom line
The organisations capturing the most value from AI are not those with the biggest budgets. Research suggests 20% of organisations are capturing 74% of AI-driven returns (AI performance study). The differentiator is strategic clarity.
CMOs who move from fragmented experimentation to a deliberate, outcome-led approach, built on small, measurable projects and the right leadership mindset, are already pulling away from the field.
