AI’s Pinocchio paradox

Nick Keenan
By Nick Keenan | 28 June 2023
Nick Keenan.

Nick Keenan, CEO, Starcom

What does media’s AI technology and the story of Pinocchio have in common? Well, like the wooden toy in the children’s fable, media AI is artificial, it’s awe inspiring, it talks, it lacks judgement and its many Geppettos (creators) are desperate for it to be real.

It is well-packaged too. As with Pinocchio being a state-of-the-art wooden imitation of a real boy, AI, armed with A Space Odyssey’s HAL-inspired Natural Language Generation (NLG) and creative automation of past knowledge, is no doubt whizz bang in appearance. But none are independently intelligent, nor is there one that can be relied upon to solve client marketing needs in the form of some magical or uniquely federated tool. At least not yet.

Despite this reality, some (not all) are being packaged up with embroidering claims of connectivity and widely holistic marketing benefits. Sadly, our industry is grossly exaggerating the current capabilities of what are, at the end of the day, ‘probability-based algorithms’ in search of the right word. I say sadly, because it leads to unwarranted scepticism at the board level when there is a lot to celebrate and embrace.

Like Pinocchio’s dad, we all wish the boy to be real. But it's a long road from wood to flesh, and the key barrier that I can see is that AI is operating within disparate platforms – disconnected across multi-channel universes, effective in their own isolations but seldom connected to get the true benefits of automation and real-time insight.

For us in media that means in strategy, planning and investment, it is still very high touch. Every level of new automation breeds another complex problem that needs connecting. The compounding problem is an ever-expanding universe in media fragmentation.

If we consider walled vs. un-walled media environments, the speed of change is now recoding 76% of all online traffic occurring in un-walled environments, according to GWI. Audiences in these locations are increasingly moving away from big technology; the scaled digital monopolies of yesteryear that were so easy to plan and buy media within. No, we didn’t get all the data, but it was an easier channel selection and plan to deliver – the results equally as good. So, with splintering audiences, big tech’s AI tools now only cover 24% of daily total traffic in an ever more complex channel mix.

In spite of these disconnected realities, the claims of AI in advertising stretch well beyond the nose of the fabled wooden boy. It seems in the media AI arms race, there are those who believe they can ‘out promise’ to win business, packaging it up as a tool ‘no one else has’ when, in reality, it is often just white labelled tech that we all have access to or a tool so isolated from daily operations it is of little practical use.

So, what role does AI play in the marketing landscape? Well, we can’t ignore the technical advances being made (almost daily), and being a technology Luddite will not end well for you, your talent or your business. It will yield better ways of working and leave those that don’t adapt behind, but what is useful right now? What should a marketing team be asking of us agencies? And, ultimately, what does success look like?

Success in applying AI is about connecting it to a customised platform based on client needs. What it is not is applying or trying to force intrusive technology that suits the intruder. A client-first approach can leverage the useable advances in AI by incorporating it into specific strategic and planning benefits for an independent client solution. AI’s influence in media planning and buying should look further upstream, focusing on its contribution to strategy and planning. These are the two fundamentals for a media agency where clients should be focused given the latest changes, not in the obvious programmatic, unit dimensional bottom funnel conversion.

AI applications in bottom funnel conversion is now table stakes in the larger agency groups and no the technology has not democratised the playing field enabling smaller operators to do more with less. Size and scale may not have its historical advantages but it still matters, especially when considering the scale of a client, their products, services, customer segments, digital maturity, and, of course, varying levels of structured data. Depth and breadth in talent and easily modularised capabilities to fit is paramount. Because there is no one-size AI shoe, but a wide array of deep domain integrations, from highly-talented and always-on trained talent. This will connect where possible the useable AI-driven media contributions, and right now that is focused on three core functions:

  1. More efficient and accurate automation;
  2. Geo-targeted predictive modelling; and
  3. Dynamic consumer insights

For this level of automation, predictive modelling and real-time audience insights to work and connect to broader strategic planning frameworks, your team and structure has to be set up to accept AI’s disparate contributions. We can’t make the AI boy real and none of it all links together simply or easily, but cultural agilities can make AI deliver more efficient and effective outputs through discerning integrations of some of its best parts. For us what is real, while not holistically connected, exists in five key disciplines:

  1. Virtual customer segmentation (virtual focus groups and audience segmentation)
  2. Near real time MMM
  3. GEO-propensity modelling
  4. Omni-channel activation in non-walled environments
  5. Programmatic investment and optimisation (this takes big, well-organised teams, but is table stakes)

The benefit of applying these five with their varying levels of AI function enable more, real-time, intimate knowledge of brands’ customer segments and their behaviors, influenced by dynamic cultural trends that are moving faster than ever. Further, they provide a contemporary media channel mix that is reflective of today’s media, delivering omni-channel implementations, and optimisations, armed with predictive planning to test, learn and increase conversions at the bottom of a vastly different digital funnel.

So, Pinocchios aside, AI is really about accepting the vast and chaotic growth in technology and being adaptive and agile enough to pick up its better made toys. Once you find them (and can afford them) it is then about implementing them into agency structures, talent training programs, clients’ owned ecosystems and the broader campaign workflow. My advice? Look for the good boy Pinocchio – the honest one that is interested in looking after what matters. The AI that wants to save us from the data whale, not run off and leave its creator stranded in its belly.


comments powered by Disqus