Why scaling and challenger brands need to use AI in marketing without outsourcing the judgement that gives them an edge.
Barely a day goes by without another lurid headline about AI’s lethal impact on people and planet. The more prosaic reality of how it is deployed inside businesses and workflows gets left in the shade.
I recently spoke on a panel at ATS London on the “tokenomics” of AI infrastructure in marketing, which turned out to be about value, governance and adoption. That conversation usually centres on the enterprise: the multibillion-pound advertiser balancing a token budget, the brand with millions of user records and almost as many lawyers. The scaling challenger brand, with far more ambition than resource, is underrepresented.
It shouldn’t be. For these businesses, value, governance and adoption are the CMO’s job as much as finance’s or legal’s. Few operators are better at demonstrating effectiveness out of uncertainty, under pressure to control cost while treating marketing as an investment in growth. That pressure pulls towards commoditisation, and it is where AI can do the most damage. Here is where I see it across planning, creative and data.
AI and media planning for scaling brands
Media planning is where commoditisation bites hardest. This isn’t an attempt to put it on an artistic pedestal, but it is far more than dropping rates and benchmarks into a spreadsheet and asking for an IO. Brands that reduce it to that are the most exposed to homogenised thinking. Considered planning weighs macro factors against brand context, with a clear measurement framework underneath. Conflate it with activation, or judge strategy on a Pareto curve, and you hand your competitive advantage to everyone else.
The same logic applies to agentic trading, the idea that agents will negotiate on behalf of buyer and seller, moving what once happened over a long lunch into ones and zeroes. The innovation is real. But scaling brands know the cost of being first mover, and risk awareness is not risk aversion. At MFN, any experiment ladders up to defined objectives and business targets. We will pioneer, but it will be verifiable.
AI and creative production
Platforms want ever more creative variations to feed their optimisation. The result is a glut of near-identical assets, most of which will never be seen, judged by an algorithm that reduces creative production to a single variable. AI tools can ease some of that pain without removing human thinking from the process, and we use them with stakeholder buy-in and human review before anything reaches a channel. Demand for more is not a reason to satisfy it.
AI, marketing data and insight
This is closest to my own patch, and where AI swings furthest between upside and downside. Data without context is just information, and value comes from turning it into intelligence. Asking an LLM to tell you what the data shows and why is one of the surest paths to cognitive atrophy. Used well, though, it lowers the barrier to things that were out of reach, such as assisted Python or R scripting that gives a credible directional read between platform vibes and a slow MMM. That only holds if there is enough domain knowledge behind it to challenge the construction and the outputs.
Where scaling brands should draw the line with AI
Across all three, the pattern is the same. AI is a good place to offload the menial and a bad place to offload judgement. Challenger brands cannot afford to confuse the two, because their edge has always been better thinking per pound spent, not more spend.
So the test for any AI use is simple. Can you verify it, can you explain it and would you defend it to your CFO? If the answer to any of those is no, the tool is doing the thinking, and you are paying for the privilege of becoming a commodity. For founders and senior marketers, adopting AI is more than a full-time job. It still takes human brains to sculpt what comes out of the GPUs.
