Agencies and AI: Ongoing Investments in People Limit Risks From Technology
Twitter’s failure to operate as intended during yesterday’s scheduled Elon Musk-Ron DeSantis Twitter Spaces event will surely be remembered for many reasons, not least as it presumably serves as a reminder of the consequences of the severe cost cuts Twitter has endured since Musk’s acquisition last year.
To me the episode highlights the sometimes under-estimated importance of people in an increasingly technology-dependent advertising industry. In recent months many pointed to “bloat” at many of the largest media platforms as justifying recent layoffs, and perhaps there was some. For those making such arguments, Twitter was held up as a symbol of the opportunities to improve profitability and reduce bureaucracy (and thus improve growth potential) that could follow from such actions. However, if it wasn’t already clear to most that Twitter went too far, perhaps it will be now.
To the extent that there was ever a presumption that better technology should mean fewer people working it also provides a useful reference point about how the services-dependent advertising agency sector – especially in the traditional “creative” advertising agencies who historically were responsible for the production of campaigns – may evolve as artificial intelligence tools proliferate. This has become an increasing concern in the minds of many investors and industry observers, as evidenced by the frequency with which related questions have been raised on recent agency holding company earnings calls.
While it’s hard to predict the future with any precision here, my take is that so long as wise decisions are made by everyone involved, outcomes are probably neutral to positive for the legacy creative agencies. My view is largely informed by the way in which automation impacted media agencies over the past twenty years.
From the turn of the century, with the rise of digital media it was clear that automation was going to profoundly affect these businesses. As a consequence, by the early 2010s there were many who thought that this automation would lead to disintermediation of media agencies, as it would become increasingly possible for marketers to do much of the work media agencies did using tools made by technology companies. Alternately, there was a view that cost pressures would push harder than ever on media agencies, which were already perceived (wrongly) as low margin businesses because many of its practitioners conflated take-rates with profits.
Although it was possible to do, few marketers who weren’t already managing their own media ever actually in-sourced media at scale because the introduction of programmatic trading required new and evolving skills among practitioners to manage relatively simple existing processes such as buying (negotiating for, stewarding and paying for purchases of ad inventory), while new campaign management techniques required new services such as data management. Moreover, the complexity of the ecosystem and the wider range of potential ways to manage campaigns made measurement more complicated, requiring additional labor. Incidentally, AI was increasingly relied upon, as buyers and sellers required higher levels of automated decision-making to manage all of the new opportunities that emerging media channels supported along with all of the data they produced.
As automation became more and more important, so too did the media agencies who were most exposed to this trend. My guess is that the holdcos’ headcount at media agencies probably grew by around 80% over the period between 2012 and 2022, roughly similar to the growth of the overall advertising industry outside of China (where agencies do have some exposure, although proportionately much less than China’s share of the global advertising industry). Commentary from industry executives about these businesses during earnings calls suggests profitability remains strong, too.
Looking forward, while there certainly are some risks to traditional creative agencies from generative AI, I’m doubtful that they will be negative over extended time horizons. Already a substantial volume of creative production occurs through specialist entities which are heavily dependent on automation techniques, including those which involve AI. Moreover, the hardest work associated with traditional advertising agencies arguably relates to the socialization of ideas among a marketer’s management teams, including the hundreds of stakeholders inside and outside of a company who need to sign off on and believe in a brand’s marketing strategy at the larger companies agencies often serve. Marketers’ management teams “own” the strategy, but traditional advertising agencies were historically heavily involved in these socialization efforts as well as part of their account management capabilities. While no agency was cited for its involvement, I was certainly reminded of the issue of idea-socialization while reading the Wall Street Journal’s deep dive on what happened at AB Inbev in the wake of the distribution of a customized Bud Light can to influencer Dylan Mulvaney.
While it’s probably true that if marketers’ requirements for creative work was unchanged from what it is in the recent past, improving generative AI capabilities would significantly hurt the agencies responsible for developing, producing and socializing the creative aspects of campaigns. But if instead practitioners focus on new and different ways to add value – using better insights or new ways to look at consumers and brands, for example – or find new ways to maximize the impact of AI-based marketing on consumers, or develop new tools that become increasingly essential to manage AI-based campaigns, it’s not hard to be at least neutral or possibly positive about the impact of AI on traditional advertising agencies. At least we can be so long as agencies impacted by AI continue to make necessary investments in the people who will ultimately drive the health of the advertising services industry.