Sir Martin – founder of what became the world’s largest advertising company, WPP – is now chairman of the board of S4Capital, which aims at digitally native delivery of services.
He talks about the use cases in advertising and marketing, limited currently, but where activities like planning and buying media will likely become automated by AI and work algorithmically.
Sir Martin also talks about how AI can bring a huge leap in productivity across all sectors and why enterprises need to take advantage of that by integrating the work of the chief information officer (CIO) and chief marketing officer (CMO) and bring data to the fore in their work.
With your experience in the sectors you work in, how is AI affecting the advertising-marketing-brands vertical?
Well, significantly, I guess. I mean, there’s a caveat with that in that the actual use cases are few and far between.
So, when you work with the leaders in the AI industry, like an Nvidia or whoever, they are very focused on what are the use cases. And when you look at use cases, I would say we’ve got a lot of tests going on, audits, workshops going on. But actually the number of clients who are willing, for whatever reason, to implement at scale are relatively limited.
But, with that as a sort of health warning, if that’s the right way of putting it, basically we see five things happening in our industry. The first is around visualisation and copywriting.
So, that’s producing ads. It takes less time. And that is a two-edged sword because our business, our industry, the industry is a trillion dollars. Clients spend about a trillion dollars on advertising, about $700bn in digital and about $300bn in traditional.
Digital is growing, probably at around 10% or 15% a year, and growing its share. So, when we started S4 six years ago, digital was probably about 50% of the market – it’s now 70%.
And traditional is declining – probably about 0% to 5% a year if you have live sport. And, if you don’t have live sport, you’re probably down 5%, 10% or 15%. And that continues to be the dynamic in the industry. So, really there are two industries.
But with that as background, a lot of that revenue is generated by agencies, by time. And, if you’re compressing the amount of time for making ads, as you are with copywriting and visualisation, what took you weeks or days is now taking you hours or minutes. The amount of time is reduced, and procurement departments, the clients – quite rightly – say, it takes you less time, it should cost you less. So I would say there’s compression. There will be fewer copywriters and fewer art directors as a result.
So, that’s one area.
The second area, which I think is really interesting from a growth point of view, is personalisation at scale. So, we work for a Netflix, or an Amazon Prime, or Disney Plus or Univision, and we can produce, if we’re launching Squid Game or Narcos for Netflix, historically we might produce about a million and a half assets, at least in theory.
We’re now producing multiples of that using AI and personalisation at scale and using first-party data and signals from the platforms. So, that’s a huge area of opportunity. The price per asset is falling, but the number of assets used is rising at such a rate that there will be more employment for people in that area.
So, that’s the second area.
The third area is media planning and buying. Unfortunately, for the holding companies that dominate the media planning and buying space – probably there are six of them of scale, which probably have about 60% between them of the market in terms of planning and buying – they employ about 200,000 people [but] there won’t be 200,000 people in two or three years’ time. Unfortunately, also, the processes are very manual.
If you compare it to the investment industry, to a BlackRock or a Blackstone, which manage more than one trillion each in terms of funds under management, they don’t manage those investments and the distribution between property or equities, or private equity, or gold, or bitcoin, real estate, or whatever it is, they don’t manage it manually or semi-manually, they manage it algorithmically. And the same thing is going to happen with media planning and buying.
And, you see that already with the platforms like Google and Meta – Google had with Pmax, Performance Max, and Meta with Advantage Plus.
Actually, it was really interesting. The CFO [chief financial officer] of Meta said in the fourth quarter earnings for last year, or this year, but commenting on last year, Meta’s total ad revenues are about 160 billion versus Google’s 260 billion and Amazon’s 61/60 billion and TikTok’s 40 billion outside mainland China. So, those four platforms are half of total ad spending – 500 billion between them – and about 70% of digital.
But the CFO of Meta said Advantage Plus accounted for about 17 or 18 billion of the 160 billion. So really 10% of Meta’s media planning and buying is being done on a program which basically delivers algorithmic answers to small and medium-sized businesses, just like Performance Max does at Google, although we don’t have the data. So, really interesting what’s happening on media planning and buying.
So, you won’t depend on a 25-year-old media planner and buyer in the future for input and output. That buyer will have a much more sophisticated output coming from algorithmically analysed planning and buying decisions.
So, that’s the third area. The fourth area is general efficiency.
I’ll give you an example of that. We have a joint venture with Nvidia and AWS [Amazon Web Services] and Adobe around outside broadcasting. If you’re doing an outside broadcast, you need a truck for that in conventional ways. And that truck will cost you 10 million [dollars], you’d amortise it over five years at two million [dollars] a year. We can provide a cloud-based solution with Nvidia, AWS and Adobe for $100,000 or $200,000.
I referred before to the traditional media companies being under heavy pressure, so you can see that they would leap at the opportunity of reducing their costs and becoming more efficient. It’s a good example of agency and client efficiency.
And the fifth area, the last one, is democratisation of knowledge.
I don’t think there’s any accident that [Nvidia CEO] Jensen Huang has 51 direct reports. If Computer Weekly went to McKinsey and said, ‘What’s the ideal organisational span or spread for a CEO?’, you would say 12 or 13. Jensen manages it apparently with 51 direct reports or thereabouts. He also doesn’t have as many one-to-one meetings.
What he tends to have is … KPIs [key performance indicators] for his direct reports and then monitors their progress. So, what you’re getting is the flattening of organisations and AI reducing the silos and the politics that go on inside big companies, and making them much more efficient and effective, and simplifying processes and bringing people together in much more effective ways.
So, those are the five areas – just to sum up: visualisation and copywriting, personalisation at scale, media planning and buying, general efficiency and democratisation of knowledge.
You’ve given a broad description of how it’s affecting your area of things, so how might AI go into other areas, such as manufacturing or logistics or whatever?
The answer is it will affect … sorry, not ‘will affect’; it is affecting everything.
Probably, to be fair, affecting things like manufacturing more so than mine because, with the exception probably of General Motors – that’s the biggest example that we’ve come across.
I mean, we have Google Hatch, we have Forever 21 with Meta, we have some work with SC Johnson and others. But with those exceptions, I think the one case study in our industry where a company has literally turned itself, its marketing model on its head and it’s doing that for competitive reasons.
We’re seeing in the auto industry and the financial services industry the rise of lower-cost, lower-price competitors. So, in the case of the auto industry, Chinese EVs [electric vehicles] and autonomous vehicles; in the case of financial services, fintech companies.
Now, these new competitors can produce products or services at lower costs and therefore the traditional manufacturers or the traditional financial services companies have to make their model much more efficient.
So, the answer is it affects other verticals deeply, and probably more fundamentally so far.
I think it will fundamentally affect advertising and marketing in the future, but I think it’s affecting things like manufacturing and distribution far more quickly.
What do enterprises need to do to take advantage of the things AI is bringing?
There are three pieces of advice we – CMOs or CIOs – would focus on, because when you talk about AI, it’s really uniting several functions.
It’s uniting marketing, it’s uniting technology and IT. I often think there should be conferences with CMOs and CIOs – those are the two.
I remember John Chambers used to run at Cisco every two years a conference for CEOs and CIOs, and I think that from an AI point of view, it’s really important to get the CMO and the CIO together because their worlds are colliding, for want of a better word, but I think there are three areas.
Agility is key. Everybody talks about agility, but delivering it is much more difficult, so that’s one thing.
I would say taking back control after the great financial crisis in 2008, people tend to outsource. You know, finance and procurement said outsource, get the headcount out of the company. So a lot of functions were delegated outside, and that may sound strange coming from somebody in the agency business, but that may be misdirection and what you have to do is to exercise more control. And the third and final point is the importance of first-party data, which is very much more focused.
So, with deprecation of third-party cookies by Google – not totally, but directionally. [And] with the change in the IDFA rules with Apple moving from individual IP addresses to cohorts, to groups. The use of first-party data – that is, consumer consented data which can be used freely, or as freely as possible – combined with the signals from the platforms, from the three western ones I mentioned before, Alphabet, Meta, Amazon, and then in the east, Alibaba, Tencent and ByteDance (or TikTok); those are the three, the six games in town really in terms of scale, using the signals from those platforms, plus the first-party data critical.
I would say those are the skills, organisational skills – agility, as I say, everybody talks about it, but it’s very difficult to deliver it, particularly in a volatile, you know, in a Trump 2.0 world, you have to be very adaptable and very agile. So, that’s one area, take back control. You’re going to have to control the data, integrate the data. Very few clients have first-party data at scale in a fully integrated way. And then finally, first-party data.
