It’s Q4 2027, and it’s time to forecast your marketing spend and return for next year.
Your marketing team - now one person, a grad you hired last year after downsizing hit - asks your in-house AI agent where to allocate your budget. It analyses all of the platform data from last year, and spits out a recommended investment split by tactic and publisher. With the fast pace we now work at with the help of AI, this recommendation shoots straight up the pipeline to the GM, and the budgets are approved within a day. Great!
But… What was this decision actually based on? Your marketing manager doesn’t truly know, and you certainly don’t.
You ask the agent why it allocated budget to those particular channels. “Based on historical performance, these channels are expected to produce the most return.” Sounds good!
Three months later, overall sales are significantly down year-on-year. What’s gone wrong? Were there faults in the original dataset? Have there been unforeseen changes in the marketing environment? Was there a bug in the agent? Did consumer sentiments suddenly change? Was there a tracking problem this year, or even last year? Where do you look first and what questions do you ask?
As Pablo Picasso said; “What good are computers? They can only give you answers.”, and just like the number “42”, an answer is worth only as much as the questions that are asked.
We’ve got the data, we’ve got the processing, but if you’re not following a thorough and rigorous line of investigation, your own personal ability to question the data itself and understand what’s actually happening, you’re being left in the dark. Data in any form - raw, cleaned, summarised, visualised - is only as valuable as the insights that can be gleaned from it. The perception of what “insights” are has unfortunately been reduced to dot points on a slide, or three sentences in an executive summary. Real insights are actionable and (importantly) account for the complexity of relationships between variables.
Forgive me for wading back into the territory of high-school science classes, but one of the most fundamental skills in business, let alone marketing, is an understanding of dependent, independent, and moderating variables. I’m not suggesting that every individual meeting needs to start with a reiteration of the null hypothesis, but rather that to actually understand how effective marketing is, one needs to be able to conceptualise the most important variables present and the relationships therein.
Can AI assist with this? Of course. Can AI upload inherent understanding of all of the factors & relationships involved in my dataset into my brain? No, it cannot - and when the questions land in my inbox, the only things that can provide that knowledge are experience and familiarity. We’ve got platforms that can produce thousands of insights, but these are just insights for insights’ sake. We’re hitting cruise control, being drip-fed a steady stream of digestible soundbites that sound useful on the surface, but without the ability to interrogate them at a deeper level, are ultimately just window dressing.
What’s more stagnant than a company on autopilot? I despair at a vision of marketing departments simply following platform output with no critical thought, of CEOs and GMs asking for contextless summaries from LLMs with no understanding, of juniors who will never develop creative solutions and novel approaches because they have no comprehension of the fundamentals of systems.
It sounds very dramatic, but: If you’re a marketer, have some pride in yourself as a marketing specialist! If you’re a leader, have some pride in your team and company! If you’re a grad or a junior, for heaven’s sake, dig into the processes and gears and levers and formulas, unmask the system! We need our people to be empowered, because only empowered people can challenge the status quo, can be truly creative and innovative, can evolve.
This requires collaborative effort from top to bottom - almost like it’s some sort of organisation! There are no shortcuts. Use technology and use it well, but without expertise at the helm, long-term growth is all but a dream.
Luke is a Co-Founder and Head of Growth at Hard Refresh. He wouldn't describe himself as anti-AI, but he is definitely pro-people. You can connect with him on LinkedIn here and see how much trouble he gets himself in.