AI can analyse data. But can it make better media planning decisions?
Every few months, another headline predicts that artificial intelligence (AI) will transform marketing. Media planning is usually next on the list. Algorithms can analyse audiences, optimise campaigns, automate media buying and even generate advertising content in seconds. With capabilities like these, it is easy to wonder whether media planning is becoming a technical problem that machines can solve better than people.
Before exploring global media planning in more depth, I probably would have agreed.
My understanding of media planning was fairly simple. I thought it was mainly about selecting the right channels, allocating budgets and reaching as many people as possible. If that was the job, then AI seemed perfectly positioned to take over. After all, machines can process far more data than any human planner and optimise decisions much faster.
Over the past few weeks, however, my understanding has changed considerably.
What surprised me most was discovering that media planning begins long before anyone chooses a channel. Through the eBay Stronger as One case study and the guest lectures from Maya Epstein and Penelope Bellegarde, I came to understand that media planning is fundamentally about decision-making. Before a planner considers television, search or social media, they must first answer much more difficult questions. What business problem is the campaign trying to solve? Who is the audience? What behaviour needs to change? How will success be measured? What role should different touchpoints play throughout the customer journey?
These questions cannot be answered by data alone. Realising this completely changed how I think about AI.
There is no doubt that AI is becoming an important part of modern media planning. It can analyse large datasets, identify audience patterns, personalise communication, optimise bids in real time and automate many routine planning tasks. McKinsey & Company (2023) argues that generative AI has the potential to transform marketing by improving productivity and supporting faster decision-making. For international campaigns involving multiple markets, platforms and audience segments, these capabilities represent a significant advantage.
These capabilities are impressive. If efficiency were the only objective, AI might eventually replace much of the planning process.
But the more I learned, the more I realised that efficiency is not the same as good judgement.
One concept that particularly challenged my assumptions was Google’s idea of the Messy Middle (Google, 2020). Rather than moving neatly from awareness to purchase, consumers constantly move between exploration and evaluation. They compare products, read reviews, search for information, ask friends for advice and often leave before returning later. Different touchpoints perform different roles throughout this journey. Some introduce a brand, others build trust, reduce uncertainty or encourage action.
Figure 1 illustrates how consumers move repeatedly between exploration and evaluation before making a purchase, rather than following a simple linear path from awareness to conversion.
This made me realise that successful media planning is not simply about reaching people. It is about understanding why people behave the way they do and deciding how communication can help at different moments in their decision-making process.
Penelope Bellegarde reinforced this insight by explaining that customer journey analysis is valuable because it identifies where customers experience barriers, lose confidence or need additional reassurance. Analytics, therefore, do not create value on their own. Their value lies in helping planners make better decisions.
AI can certainly support this process. It can detect patterns in customer behaviour, identify where consumers abandon their journey and recommend improvements based on historical performance. These capabilities make planning faster and more efficient.
Get Arsène Birindwa’s stories in your inbox
Join Medium for free to get updates from this writer.
However, these capabilities also reveal an important limitation. AI is exceptionally good at analysing information and answering questions, but it is much less capable of deciding which questions matter most. That responsibility still belongs to the planner. Figure 2 illustrates how AI supports media planning by analysing data and generating recommendations, while human judgement remains essential for interpreting insights, applying context and making strategic decisions.
Reading the eBay Stronger as One case study made me realise that the campaign’s success was never just about choosing the right channels. It began with a strategic decision: to support small businesses rather than promote the brand itself. During the COVID-19 pandemic, eBay invested £9 million in promoting the stories of small businesses across television, print, radio, outdoor advertising and digital media. The campaign positioned eBay as a genuine partner to small businesses during an exceptionally difficult period and resulted in a 256% increase in new businesses joining the platform, alongside significant sales growth for participating sellers (Jordan, 2020).
AI could almost certainly have optimised the campaign’s media mix. It could have recommended budget allocations, identified the most efficient channels and monitored campaign performance in real time. However, I find it difficult to imagine AI proposing the campaign’s central idea. Choosing to support small businesses rather than promote the brand required an understanding of public sentiment, social context and long-term brand positioning. Those decisions involved judgement rather than optimisation.
This seems even more important in global media planning.
Consumer behaviour differs across countries. Platform use varies. Privacy regulations evolve. Levels of digital access remain unequal. Cultural expectations influence how messages are interpreted. The same AI recommendation may perform exceptionally well in one market and fail completely in another. Data can reveal patterns, but it cannot always explain why those patterns exist or whether they should shape strategic decisions.
Looking back, I realise that AI was the reason I chose this topic. Ironically, it is no longer the most interesting part of the discussion.
What interests me now is decision-making.
Media planning is not simply about processing information more efficiently. It is about making informed decisions under conditions of uncertainty. AI is becoming exceptionally good at analysing information, but analysis is only one stage of the planning process. Someone must still define the problem, interpret the evidence, balance competing objectives and ensure that planning decisions are commercially effective, culturally appropriate and ethically responsible.
My conclusion is very different from the one I would have reached at the beginning of this module. I no longer believe AI will replace media planners. Instead, I believe it will redefine what makes them valuable. As automation takes over routine analysis and optimisation, the planner’s greatest contribution will increasingly lie in asking better questions, interpreting evidence critically and applying judgement where data alone is insufficient.
The question is no longer whether AI can analyse data better than people.
It probably can.
The more important question is whether AI can understand people, culture and context well enough to make the decisions that matter most in global media planning.
That is the question I explore in the next post.
Further Reading
If you’d like to explore the ideas discussed in this article, these resources provide an excellent starting point:
Google (2020) — Decoding Decisions: Making Sense of the Messy Middle
https://business.google.com/uk/think/consumer-insights/navigating-purchase-behavior-and-decision-making/
Jordan, L. (2020) eBay: Stronger as One. WARC Media Awards, Effective Channel Integration. London: WARC.
McKinsey & Company (2023) — The Economic Potential of Generative AI: The Next Productivity Frontier
https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-economic-potential-of-generative-ai-the-next-productivity-frontier
