AI in Marketing
Join over 3,500 marketers, operators and investors that read The Raffi Report by subscribing here:
To the regular readers, if you enjoyed today’s piece please share, comment and subscribe!
AI has escaped from academia. Now firmly in the public consciousness, the stakes for all sectors, including Marketing, are taking shape.
To date, anecdotal discussion sets the future up as a binary. People explain either how AI tools will give superpowers to Marketing departments, or how these tools will make marketers redundant en masse. Tweets abound with a thread of 100 AI tools, still all MVPs, that are “going to change the entire world”, while others cry some version of “everyone is going to lose their jobs!!!”.
Three difficulties are inherent to this framing.
First, it treats Marketing as homogenous, rather than a category made up by a range of specialisms. Second, it presupposes a fixed level of expertise with the tools, when all technology presents some degree of learning curve. Third, it does nothing to specify which people will face redundancy.
On this last point, I draw your attention to the reminder
provided in his recent interview with Cal Newport:A few years ago I was looking at news coverage from the early 1970s when the ATM was introduced. Some of the coverage was apocalyptic — 300,000 bank tellers are going to be out of work overnight! But instead, over the next 50 years, as there were more ATMs, there were more bank tellers. ATMs made branches cheaper to operate, so banks opened more branches. Fewer tellers per branch, but more tellers overall. But more than that, it fundamentally changed the job, from one of repetitive cash transactions, to one where the person is, say, a customer service rep, a marketing professional, a financial adviser, etc. They needed a much broader mix of more strategic skills to add value.
— Inside the "Mind" of ChatGPT,
, April 2023
In the same vein, my colleague Andrei Bujor-Blanaru made a neat comparison with the Industrial Revolution; which ended the careers of many generalist workers, but permitted strategically-focussed and specialised ones to flourish, such as blacksmiths and bread-makers.
These historical examples overwhelmingly inform my perspective. The objectives and therefore needs of Marketing teams are going to change, and so the people required to deliver those needs will change too.
Does that feel apocalyptic? I don’t think so. But it is important we keep track of who is enabled to use AI effectively. If we take AI to be a new kind of electric grid that powers everything, I wonder which people will be the first to acquire adaptors and become compatible with it? Unless we are thoughtful in our approach to ensure that as many people as possible can use it proficiently, we face a material risk that AI fortifies existing structures of inequality and skills gaps in the Marketing industry.
What do we mean by AI?
Important to this debate is the distinction between Generative AI’s capabilities as a knowledge base (a narrow tool that can inform future decisions based on data collected in the past, or “rules of the system”) and its potential to serve as an intelligent actor; in which it could play co-pilot to humans in situations in which it has no experience.
This intelligent actor, otherwise known as Artificial General Intelligence (AGI), is a further frontier not yet reached. As ChatGPT explains it:
Safe to say, the “if” and the “when” of AGI are contested by folks much smarter than myself. (Forced to choose, my vote lies with Ian Hogarth from Plural, who evasively proposes that “credible estimates [for the arrival of AGI] range from a decade to half a century or more.”)
Upon reaching AGI, however, with systems that are disobedient, then all bets are off for all white-collar workers of every stripe; marketers, engineers, lawyers and the like.
So for now, I will assume we remain outside the realm of AGI, but in a realm where Generative AI’s power, albeit more narrowly useful, continues to improve exponentially. In that realm, what happens to Marketing?
Personalisation
Today, one of the clearest blind spots in Marketing, and specifically digital channels, is the promise of platforms to target consumers accurately, and the experiences of consumers that receive the adverts.
At present, even if third-party data could give us a complete picture of a consumer, iOS 14.5 has meant that platforms are often unable to deliver on the fullness of that data.
Generative AI has the potential to rectify some of that fault. In the words of Sir Martin Sorrell:
“What AI will enable us to do is to produce hundreds of different versions of the same ad that are based upon everything we know about an individual consumer: the whole data set. You will get an ad that is much more “empathetic” to you and it should feel a lot more relevant. There will be new formats: you might be served an ad on your mobile phone that actually shows you wearing the shirt or the watch being advertised.”
— Why the Mad Men would be crazy to ignore AI, The Times, 2023
Personalisation is a contentious topic in Marketing. For Peter Weinberg and Jon Lombardo of the LinkedIn B2B Marketing Institute, “one-to-one personalisation at scale” should be considered “the worst idea in the marketing industry”. In their assessment, the case against personalisation “can be reduced to two simple words: 1. couldn’t, 2. shouldn’t.”
In the case of “couldn’t” - Weinberg and Lombardo present analysis from teams at MIT, Melbourne Business School and HP to show that in a B2C context, targeting is accurate between 4% and 44%, while in B2B it’s between 7.5-14.3% dependent on the seniority of the stakeholder. In other words, prior to 2022, targeting was less accurate than a coin flip; despite Meta breaking EU laws to harvest personal data.
Now, thanks to AI, the quality of targeting looks like it is on the rise.
In paid social, early evidence from the introduction of Meta’s Advantage+ program, which promises to eliminate “the manual steps of ad creation and automates up to 150 creative combinations at once [to help] advertisers more quickly learn what ads are working, while making the most of their advertising budget”, is a fulfilment of that promise, even if the trade-off for brand teams is a vast reduction in control. In paid search, Google is now incorporating its Generative AI, Bard, into its Performance Max program. Of course, we are still early, but logically if AI can test creative iterations and learn at such speed, then personalisation and relevancy will improve.
And yet, even if there is a scenario in which AI eliminates the first charge of “couldn’t”, to me there is no apparent progress on the second charge of “wouldn’t”.
The open question is not whether AI can deliver personalisation, but whether it can deliver effectiveness through personalisation over and above this concept of performance branding.
To wrestle with Sorrell’s idea, would it not be creepy to see yourself in a shirt in an Instagram ad? I like the idea of AI testing which celebrity influencer should be wearing the shirt, but we’re still not ready to disprove Weinberg and Lombardo in their argument that “marketers would be much better off investing in ‘performance branding’; in other words, one-size-fits-most creative that speaks to the common category needs of all potential buyers, all the time.”
Of course, time will provide fertile ground for brands to discover the answer, and doubtless, following on from Nestle’s recent announcement, there is going to be substantial investment in this space.
But there is a high bar. Consider the Effie award-winning “Moldy Whopper” campaign by Burger King from 2022, which delivered a 26% uplift in quality ingredient perception, a 23% rise in consideration and a 14% increase in Whopper sales on 2.9B total impressions.
Even if it might be able to produce and test certain asset permutations (for instance, showing a Veggie Royale to vegetarian consumers, instead of a Whopper), conceptually, it is still unthinkable that digital ads, however well personalised through Generative AI, could conceive of that entire campaign, or better the results it delivered for Burger King.
Original thinking and taste
This delta between strategy and execution is critical to our discussion. Be it creating a Drake x The Weeknd mash-up that goes viral, unknowingly winning a photography prize or generating a website from a sketch on the back of a napkin - the limits of an unmistakably and uniquely human creative output are shrinking. Like in engineering where we now see coding through natural language programming, or in architecture where DALL-E 2 is generating mock-ups of buildings, AI’s usage in Marketing is going to decrease the time expenditure and experience required to execute a given result.
Let’s be clear though: none of that progress will eliminate the value associated with knowing whether the conceived output is the “right” one. In each of these cases (the simulated song, photo, or website), only human subjectivity can strategically judge the value of the output. In the words of DJ-cum-marketing-guru David Guetta, "nothing is going to replace taste", we simply are “going to use all these modern instruments” to express that taste. If anything, with 30m+ new pieces of art being created daily through AI, that saturation will need even more curation.
In this sense, far from rendering humans useless, AI rewards our most tasteful visions. This should come as exciting news to any marketer who dares to imagine unexplored terrain, and serve as an existential risk to any marketer who is contented to play it safe. Like mid-19th century painters who began to use photography to generate references at scale, which freed them to experiment creatively and produce what we call “Modern Art”, marketers will be able to use AI to generate countless inspirational reference points, and then experiment creatively to bring us into an era of Modern Marketing.
All of this will be happy news to consumers, who will surely prefer the surprise and delight of newness. But it also has positive implications for CMOs.
The most common whisper I hear from those in Marketing C-Suites is some version of this: “people on my team don’t push the envelope”. Especially in larger organisations, you will find many senior marketers that feel their department is weighed down by fear of not getting it wrong, rather than propelled forward by a desire to get it really right. Doubtless this results from incentive structures and loss aversion inherent to big businesses in which, as Daniel Kahnemann has described it, C-Suite execs “consider each investment in the context of a greater portfolio, [while] managers essentially bet their careers on every investment they make—even if outcomes are negligible to the corporation as a whole.”
Many will say that AI technology will strip marketers of their jobs, but from my vantage point, we should be saying that AI will strip jobs of their marketers. If, right now, per Gary Vanderchuk, “most marketers don’t believe in their work; they just don’t want to get fired”, are we not heading to a more exciting future?
Those bank tellers who once endlessly and dutifully performed the most repetitive tasks, finally had to engage their brains. That new strategy brought about new growth for the businesses they served, and in turn created more roles for bank tellers. The same potential exists in Marketing. More thoughtful approaches to growth, should lead to more growth for businesses, or at least more businesses that grow. And so, there’s a world in which not only does the quality of Marketing improve, but the quantity of marketers actually grows too; a modern equivalent of the tellers at the new bank branches. In that sense, AI may well deliver abundancy over redundancy.
“Technology is a resource-liberating mechanism. It can make the once scarce the now abundant.”
Peter H. Diamandis, Abundance: The Future Is Better Than You Think, 2012
For that to happen, everything hinges on a marketer’s ability to lean into the change and switch their focus from the “how?” towards the “why?”. As we edge towards tools that help us summon up any campaign or initiative imaginable, in less time, the question is: why that one?
We can all prompt but can we all prompt?
The shift to “why” only works if we are able to sufficiently speak to customers and understand how to create products and brands that appeal to the changing psychographic and demographics needs of future generations. In this vein, diversity of perspective in Marketing teams has never been more important and will only become more so.
Removing the need for prior technical experience, familial connections into junior Marketing roles, and expensive hardware theoretically brings the less well-off, less well-networked and less well-located into play. In practice, however, as software and tooling becomes more advanced and complex, any potential progress here is at risk. So before we get carried away with where Marketing is directionally headed thanks to AI - a future where everyone who wants to be can be a lot more proficient - it is important to note that not everyone will necessarily reach the same end point.
Underpinning an individual marketer’s final destination is their level of competence with these tools. It is logical that companies like Midjourney will continue to stay open-source on Discord and lower the barriers to usage, but fluency across their UX/UI will remain too high a bar for many without adequate educational support. Consider another modern MarTech tool, Hubspot, which gave rich powers of segmentation to people who previously flyered every house or business on a street, and you quickly realise that its widespread availability has not translated into widespread capability.
If we contend that Midjourney, Synthesia, or Albert (and the rest!) is on some level just another MarTech tool, it follows that not everyone will be able to use it equally as well, and that new inabilities will be born as a result.
Read self-reported surveys and you will see that “AI is used in marketing by two thirds of B2B orgs”, but how many of those orgs are using these tools to the fullest extent of their powers? How many people inside those teams feel like they know how to use the tooling? Let alone how many people outside those teams… To quote Melissa Valentine, sabbatical scholar at Stanford’s Human-Centred AI Institute, “AI […] create[s] a new type of expertise.”
When we talk about expertise in Marketing, we mean a specific set of earned growth secrets that equips someone to deliver future growth. Those earned secrets take time to acquire, and time is unfortunately not the preserve of everyone.
recently wrote an article for , which concluded that, "our workforce will be transformed and shaped by its evolution. We should all be open to evolving, too." Absolutely, let's all be open to it, but let's also be realistic about which groups of people likely have the time to acquire such expertise. Few would argue against the importance of embracing AI, yet many are not in a position to do so. Families with two working parents, single parents who do all the household chores themselves, students learning by day and working by night to pay bills, people who are still made to commute from the surburbs into offices... These are not folks with hours spare to deep-dive into the world of AI on Twitter threads, Substack pieces or Midjourney tutorials.These privilege gaps intersect across multiple vectors, not least race, geography, income and education. As if it weren’t enough already, AI means that future is further stacked towards those with time to learn how to get ahead. It puts those with time on a square with a Ladder and those without time on one with a Snake.
Fashionable as it is to say that, “marketers will not be replaced by AI, but marketers will be replaced by AI-enabled marketers” - I ask, in that paradigm, which groups of people are disproportionately replaced? While YouTube and TikTok will carry and distribute a share of the pedagogic load, as AI’s prominence grows, we must look at how skills gaps exacerbate existing lines of inequality.
To avoid a repeat of a scenario, for instance, where practically the only place to learn Mandarin as a child is in restrictively expensive private schools, due thought must be given to who is able to access the best education in AI. If learning follows existing trends in STEM, and we fail to spread the vernacular of AI widely enough, there is a risk that rather than eliminate barriers to participation in Marketing, AI fortifies them.
For what its worth, this railing against the game of Snakes and Ladders introduced by Generative AI tooling does not come from a sense of worthiness or righteousness. Rather, to fulfill its central promise of delivering more efficiency and therefore profitable business growth, in line with with Karen Blackett OBE has shown for the marketing industry as a whole, we need lots of different people with lots of diffferent backgrounds to use it. Put simply, with or without AI, diverse perspectives in Marketing teams will still deliver superior bottom-line outcomes. So for the sake of stock prices, not sanctimony, we must not let AI undo some of the slim but meaningful progress Marketing has made towards the inclusion of a wider range of backgrounds.
Going forward: high-risk, high-reward
“The invention of the ship was also the invention of the shipwreck.”
Paul Virilio, Speed and Politics, 1977
In Marketing, AI’s core promise is to make personalisation possible (even if the effectiveness debate remains open) and bring about a new generation of Modern marketers that are creatively unshackled. Moving from a focus on “how” to “why” should leave us all excited. But truly we must remain vigilant towards the fall-out from this progress.
We must steadfastly question who is properly enabled to use these tools. Absolutely let’s breathe a sigh of relief that Marketing departments can reward experimentation and imagination over regret minimisation, but let’s not pretend that everyone stands to benefit the same amount. The wider we spread the know-how, the better the business results. Our economic progress hinges on who is plugged in to this new world, and who is left behind.
All of this, of course, until AGI arrives!
Thanks to Michael Stothard and Andrei Bujor-Blanaru for reading a draft.