AI made creative abundant. Now marketers are rediscovering quality

Kevin Simonson, CEO of performance agency adMixt, recently gave one of his clients some counterintuitive advice: drastically reduce the number of ads it runs on Facebook and Instagram.

The client, a beauty brand, was running more than 1,000 distinct ads a month. AdMixt slashed that to 200 and performance got “way better,” Simonson said. The brand’s customer acquisition cost fell 25%.

For years, many brands have embraced a “more is more” strategy, flooding social platforms and ad marketplaces with thousands of different creative iterations and letting algorithms optimize toward the best performers. Generative AI has supercharged that approach by drastically reducing the cost of creative production.

But instead of making creativity less valuable, agencies say AI may be having the opposite effect. As AI makes it possible to produce near-limitless content, endless variations can deliver diminishing returns. Instead, marketers are putting greater emphasis on genuinely differentiated ideas — and on the credibility and context surrounding them.

“This whole idea of creating a million variants of an ad is bad for brand building,” said Andrew Lipsman, the independent media analyst at Media, Ads + Commerce. “The cultural relevance you create is not through personalization; it’s about having a common cultural reference point that everybody understands.”

Simonson pointed to the April 2021 introduction of iOS 14.5, the Apple iPhone operating system, which made it easier for iPhone users to opt out of cross-app tracking, as the start of the quantity over quality era of modern advertising.

The new privacy setting disrupted the advertising industry, making it harder for marketers to target and measure ads as large numbers of users declined cross-app tracking. Meta responded by encouraging brands to embrace creative diversification, feeding its automated campaigns more creative assets and allowing its system to optimize toward the strongest performers. “It went from, ‘You need dozens of ads a month,’ to ‘hundreds a month,’” Simonson said.

That strategy worked for several years, but performance began to plateau for many brands, Simonson added. Some responded by stuffing even more distinct creative assets into their ad sets, in excess of 1,000 a month in some cases, but Simonson found this was delivering marginal returns.

“Brands do need lots of ads, but the pendulum swung way too far,” Simonson said of brands inundating ad platforms with creative assets. “It’s much easier to get people to spend more on your platform by telling them to invest in more creative.”

He now advises brands to make fewer, more calculated bets on their creative. “Meta is able to distinguish between ads easier when there are fewer of them,” Simonson said.

Lipsman said even a couple hundred ads is too many. He calls for brands to scale back their creative production to only a dozen assets or less per ad set. “If you’re creating only a select handful of variants, there is going to be more substantial differentiation,” he said. “It’s going to be a better test of what’s resonating with the consumer versus slight AI-driven variation.”

AI is accelerating this shift toward quality over quantity, both through generative tools and the AI engines platforms use to create, select and optimize ads.

Jack Johnson, vice president of social innovation and growth at Tinuiti, a performance agency, said the shift to quality over quantity has been accentuated by Meta’s increasingly AI-driven ad delivery system. Andromeda helps Meta select relevant ads from a large pool of potential ads, while its Generative Ads Recommendation Model (GEM) learns from ad creative and user behavior to improve recommendations across its advertising system.

“In the past, marketers could A/B test a large volume of creative assets that had only slight differences,” Johnson said. “In today's environment, if content pieces are viewed as similar, the algorithm will consolidate them under a single entity ID.”

In practice, Johnson said, that means advertisers may gain little from producing multiple ads that are only slight variations of one another. If a brand has five ads but four of them are substantially similar, he said Meta’s system may effectively treat those four as redundant rather than four meaningfully different creative options. “If you closely monitor diversification and fatigue scores across your content on Meta, you can certainly achieve more scale with fewer pieces of content,” he said.

Mike Margolin, a former EVP at RPA who now runs his own marketing shop, Margolin + Systems, said the shift was also being driven by platforms getting better at distinguishing relevance from engagement. Likes, clicks and other engagement metrics, he argued, are unreliable indicators of an ad's effectiveness. 

"The platforms' own systems are getting better at judging whether an ad actually matches intent rather than just chasing whatever gets clicked," Margolin said. 

Margolin pointed to a research paper published earlier this year by Walmart's ad science team that found relevance is a better predictor of ad performance than engagement. "When the platforms themselves demote engagement as a training signal, pumping out fifty near-identical variants solves a problem the platforms have already solved," he said.

The implications extend beyond Meta. Adam Ortman, founder and CEO of agency Kinetic319, expected the proliferation of AI-generated advertising to eventually make distinctive creative more valuable precisely because so much AI output gravitates toward the average.

“AI does a very good job of meeting in the middle, which means it does a great job of just providing the average,” he told The Current. The result, he predicted, will be an “ocean of sameness,” in which originality becomes increasingly scarce.

“The competitive advantage is going to come when you are bringing that newness,” he said. “You’re going to be bringing that art into creative.”

As consumers increasingly turn to AI chatbots for information and product discovery, brands are looking beyond traditional SEO and looking for ways to surface in chatbot responses, a practice often referred to as generative engine optimization (GEO).

Unlike the SEO era, when brands could produce reams of blog posts crammed with SEO-friendly buzzwords, success in GEO depends on content that is accurate, detailed and actionable.

“Long-tail content gets rewarded by GEO — the stuff that’s past the proverbial third page of results in Google, but could answer very specific questions for LLMs,” David Berkowitz, founder of the AI Marketers Guild, told The Current.

Berkowitz used the example of a consumer looking to buy a lawn mower. On Google search, a consumer might simply search “lawn mower,” whereas with an LLM, searches tend to be more specific, such as “I want a lawn mower that’s electric powered and won’t scare my dog.”

“The lawn mower brand that creates this structured content on why their lawn mowers are pet-friendly, that company might be tiny compared to brands that sell through Home Depot and Walmart,” Berkowitz said. “But they created content precisely for longtail queries, and now a 15-word query can match up against the content especially well.”

The growing premium on quality doesn’t stop with the creative itself. Ortman argued that the environment in which consumers encounter an ad also contributes to how they perceive the brand behind it.

“Your creative does have a projection of you as a brand,” Ortman said. “And that also translates into the inventory or the placements in which those ads live.”

For media buyers, that makes creative quality and media quality increasingly difficult to separate: Research suggests the environment in which an ad appears can materially affect its impact, with the value of distinctive creative depending partly on the context in which consumers encounter it.

In an era of AI abundance, then, the advantage may not belong to the marketers that produce the most. As the cost of creating another ad, image or piece of content approaches zero, what remains scarce is the ability to stand out — through distinctive ideas, useful information and the environments in which brands choose to appear.