In this special, bonus extra episode of the Data Malarkey Podcast, Master Data Storyteller, Sam Knowles, speaks to Gareth Crew, global social and digital marketing leader at Canon Europe, and John Brasington, research and analytics practice leader at communications agency Golin Ketchum, about data, AI, creativity, and decision-making in marketing communications.
And now, live on stage …
Recorded live at the Meltwater Summit on Tour: London conference on 1 October 2026, this special 80th episode of Data Malarkey explores how organisations – client and agency side – are navigating the increasingly fragmented media environment. It also assesses whether the growing abundance of data and AI tools is actually helping people make better decisions.
Tools not vibes
The conversation starts with a brilliant and memorable phrase from Gareth that captures the heart of the discussion: “Tools, not vibes.” For Gareth – whose career has taken him from print journalism to digital marketing – creative judgement matters, but it needs to be supported by evidence.
Social media provides something traditional advertising rarely offers: immediate feedback about what works, what doesn’t, and what deserves another look. The most carefully crafted campaign can underperform a hastily produced meme. Data helps distinguish personal preferences from audience behaviour.
The power of human understanding
John agrees, with an important qualification. Data has limitations, and deep human understanding of audiences and communities remains essential. Neither blindly following numbers nor asking analysts to validate predetermined conclusions produces meaningful insight.
The challenge is combining analytical rigour with creative judgement. That challenge becomes more complicated as AI reshapes how people discover information and form opinions.
Who controls the narrative?
Both guests question whether organisations can genuinely control their narratives in a world of misinformation, fragmented media, and AI-generated answers. John argues that generative engine optimisation (GEO) is becoming a boardroom concern, extending well beyond marketing into how investors, employees, regulators, and other stakeholders encounter and understand organisations.
The discussion then turns to AI’s benefits and unintended consequences.
Gareth values AI’s ability to aggregate information, accelerate analysis, and remove administrative burdens. But he worries about cognitive offloading: the tendency to outsource thinking alongside tasks, potentially weakening the skills people need to interpret evidence and exercise judgement.
Email as an “act of violence”
Meantime, John offers a description – just as memorable as “tools, not vibes” – of poorly considered AI-generated communication: “email as an act of violence”. When someone delegates writing a brief to AI without properly considering its meaning, the recipient can spend considerable time interpreting instructions that nobody has carefully thought through.
There is also a question of expertise. John finds AI particularly useful when accelerating tasks he already understands, because he can evaluate the results. Using it in unfamiliar territory creates greater risks, particularly when confident-sounding answers conceal mistakes.
So, are humans still in the lead, or merely in the loop?
Gareth argues firmly for human leadership, while John highlights the growing pressure to produce more work with fewer people. Both recognise the need for better tools without surrendering responsibility for decisions.
Perhaps the most revealing moment comes when Gareth admits that, despite his enthusiasm for evidence, organisational decisions still ultimately depend on human judgement. The central question remains whether AI is making that judgement better or simply making it easier to avoid thinking.
Tools matter. So do vibes. But neither replaces the responsibility to understand the evidence, challenge assumptions, and decide what to do next.
—
The first draft of this blog was written by ChatGPT, using a transcript of the episode and an ever-refined prompt. It was then edited by real people and tested on humans for irritancy.