AI IS TURNING THE AD BUSINESS UPSIDE DOWN / THE ECONOMIST

Creative destruction

AI is turning the ad business upside down

At the industry’s annual event in Cannes, executives are feeling the heat

The advertising business sells nothing so well as itself. 

This talent was on full display at the industry’s annual awards in Cannes, which began on June 16th. 

Advertising “not only pushes creative boundaries but also demonstrates the tangible impact creativity can have on business and society,” declared one festival bigwig. 

Trophies were solemnly awarded to campaigns promoting beer, chocolate and the New Zealand Herpes Foundation.

Yet an industry that prizes human creativity has found itself on the front lines of disruption by artificial intelligence (AI). 

Mark Zuckerberg, boss of Meta, has promised that brands will soon be able to “tell us what objective they’re trying to achieve…how much they’re willing to pay for each result, and then we just do the rest.” 

Sam Altman of OpenAI believes that “95% of what marketers use agencies, strategists and creative professionals for today will easily, nearly instantly and at almost no cost be handled by the AI.”

Technology has upended the ad business once before this century, as advertising dollars flowed from television, newspapers and other “legacy” media to digital upstarts like Google and Meta. 

As AI muscles into the creative part of the industry, which once considered itself immune from automation, a new revolution is under way. 

Who will emerge on top—and can the admen of Cannes keep the rosé flowing?

The yacht parties and poolside soirées suggest that the advertising market is in good health. 

Global ad spending has been rising at a steady clip and will grow by about 6% a year for the next three years, forecasts WPP Media, a big ad buyer.

Yet the growth is highly concentrated, and becoming more so. 

The four largest sellers of ads—Google, Meta, ByteDance and Amazon—accounted for more than half the global market last year, up from just over a third in 2019 (see chart 1). 

Eight of the ten largest sellers of ads are tech firms. 

In Cannes, the digital overlords established a tent city along the beach, spilling onto the Croisette like a smart-casual invasion force.

AI seems to be strengthening their grip. 

Machine learning improves ads’ targeting: Meta’s AI-powered Advantage+ system claims to increase brands’ return on ad spending by 22%, while Google’s Performance Max product raises sales by over 10%. 

Google has begun experimenting with ads in its AI-powered search mode, which it says could be still more effective. 

Users’ AI queries are two to three times longer than ordinary search queries, meaning the firm knows not just what people are looking for, but why. 

For brands placing ads, “AI has the most potential to improve business outcomes that they’ve seen in the last 20 years,” believes Sean Downey, president for the Americas at Google.

Other techies are deploying AI to better measure the effectiveness of ads. 

Alembic, a startup, uses contact-tracing algorithms developed during the covid-19 pandemic to track whether consumers who are exposed to ads go on to buy the product. 

Using billions of rows of data, culled from TV streams, website visits and more, it found that Delta Airlines’ sponsorship of medal ceremonies at last year’s Olympic games delivered a better return on investment than any of its other promotions.

Having conquered the market for targeting and measuring ads, tech firms now believe that they can create them—a job previously left to ad agencies. 

In Cannes, TikTok unveiled AI-powered tools that generate video commercials from text or photos. 

Meta launched a volley of new AI features, including the automated personalisation of ad copy for different audiences, in ten languages. 

Earlier this month Kalshi, a betting company, aired a cinematic, special-effects-filled TV ad made with Google’s AI tools. 

The spot was created in 48 hours, for $2,000, by a director who boasted that “I got to stay in my underwear for the entire shoot.”

Ad executives are adamant that they will not be replaced by engineers in underpants. 

AI models, designed to cough up the most likely answer, are not suited to producing arresting campaigns. 

In a world where every company can create passable ads, it will be more important than ever to commission content that stands out, agencies argue. 

Meta and TikTok’s DIY tools seem to be aimed more at small businesses than at the big brands that use agencies.

Some agencies have switched from hourly billing to flat fees in order to protect their revenue, as AI reduces the number of hours needed to put together a campaign. 

Clients may pay less than before for ad creation, but they may reinvest the savings with the agency: one marketing boss of a global brand says she is simply buying more ads with the money saved on the creative side. 

Marketers, like all managers, want to protect their budgets—“and to keep coming here”, says one Cannes regular, over petits fours at a beachfront hotel.

Investors are not convinced. 

Four of the five big agency holding companies have seen their share prices drop since the beginning of last year (see chart 2). 

WPP is looking for a new chief executive. 

Omnicom and Interpublic plan to merge by the end of the year to save money. 

Publicis has avoided the slump largely by persuading the market that it has invested enough in AI to be protected.

By the bots, for the bots

AI may soon change the ad business in even stranger ways. 

As shoppers take their queries from search engines to AI-powered chatbots, advertisers must work out how to ensure their brands are recommended by the likes of ChatGPT or Google’s Gemini. 

Some are already creating promotional material that is aimed not at humans, but at the large language models (LLMs) crawling the web for information.

Evertune, another startup, analyses brands’ cachet with chatbots by posing tens of thousands of queries about product categories. 

It finds that, when asked to recommend a brand of bedsheet, for example, the main AI models agree on the top three but then vary widely (Meta’s bots like a make called Mellanni; Casper is popular with most models, except Gemini). 

The variations are explained by training data. 

Whereas Google and ChatGPT license content from Reddit, an online forum, Meta’s model leans heavily on material from its own social-media platforms.

Influencing LLMs means influencing their sources. 

That involves building a loyal online following. 

To the relief of the big agency holding companies, it also means an enduring role for old-school public relations. 

“PR is one of the biggest levers you can pull,” says Brian Stempeck, Evertune’s boss. 

But there are other tricks, too. 

Models may struggle to read infographics, which should be translated to text. 

And they lap up detailed product descriptions of a sort that would send most people to sleep. 

A lengthy passage on a breakfast cereal’s ingredients may never be read by human eyes, but could help persuade an LLM to recommend the product to its users.

As AI-powered “agents” are increasingly deployed to carry out tasks on behalf of humans, advertisers will also need to work out how to influence these robot servants. 

Google recently unveiled a shopping agent that users can instruct to tell them when the price of a product falls below a certain level. 

Such agents may soon be given discretion to make more complex buying decisions. 

Sellers of online ads are already debating how much they should charge when these are seen by an agent, as opposed to a human pair of eyes. 

The future may be one in which AI creates ads, targets them—and then reads them too. 

The admen of Cannes should enjoy the party while it lasts.