TikTok report says AI creative needs better signals

WARC and LIONS Advisory, in partnership with TikTok, have published research on the use of community signals in AI-driven marketing creative. The study found a gap between rising AI output and perceived improvements in creative quality.

The research surveyed 400 marketers in the UK, US, Australia and Brazil involved in decisions about how marketing creative and content are produced. It found that 90% said AI had quickly become part of the creative toolkit, while 88% said generative AI had increased creative volume.

Yet fewer than half, 45%, said the technology had significantly improved quality. The findings point to a broader problem in how many marketing teams use AI tools, with the report arguing that the inputs given to those systems often remain narrow.

Demographic data remains the most common input when prompting AI, cited by 67% of marketers. That is despite 59% agreeing that traditional demographic segmentation is no longer effective, according to WARC's Marketer's Toolkit 2026 survey, which the research references.

Only 17% of marketers said they always incorporate community or audience insights into generative AI workflows. At the same time, 86% said audience behaviour and community signals will have more influence on creative development over the next three years.

Behaviour signals

The report argues that participatory digital platforms create a richer picture of consumer interests because people search, comment, share, remix, create and buy in the same environment. It describes this as an "Intelligence Loop", in which audience participation creates signals, signals reveal demand, demand shapes creative, and creative drives further participation.

The report presents that cycle as a way for brands to move from reacting to trends after they emerge to using live signals to inform briefs and creative decisions earlier. Its strategic value, it argues, lies in feeding those signals back into future campaigns and broader brand and media decisions.

Lexi Wolf, Head of Thought Leadership, LIONS Advisory, highlighted the central tension identified by the research.

"WARC's Marketer's Toolkit 2026 found that belief in the effectiveness of demographic segmentation is waning. Yet this study with TikTok found that demographics were still the number one input marketers used to brief AI. Meanwhile, participatory media environments are giving us something far richer: real-time signals from people who search, comment, share, remix, create and buy.

The opportunity isn't to make the existing system faster. It's to build a better one: one that helps brands learn from people more continuously, respond with greater speed and relevance, and turn efficiency into effectiveness over time," Wolf said.

Andy Yang, Global Head of Creative & Brand Ads, TikTok, said the research suggested many brands were using advanced tools with weak source material.

"The brands winning today are not the ones using AI to generate the most content. They are the ones learning the fastest from the people they serve. We call it cultural intelligence, and it is fast becoming advertisers' most durable competitive advantage.

Yet the research for this report shows that most brands are briefing powerful AI creative tools with weak inputs: static demographics and legacy assumptions, resulting in creative that scales efficiently but fails to connect. The future belongs to brands that close the loop by creating alongside culture, not behind it," Yang said.

Framework for use

The report also introduces a five-part framework called S.C.A.L.E. It outlines the steps the authors say marketers should follow when applying AI to creative development, from setting media and audience objectives before briefing the system to using measurement as an ongoing learning process.

The first step, Select, focuses on aligning media and audience objectives before the AI brief is written, then using platform signals to shape output. Connect calls for brands to treat creators as sources of intelligence rather than simply as distribution channels.

Anchor centres on inputting distinctive brand assets to reduce generic output. Lead addresses internal governance and transparency around generative AI use. Evolve advises brands to treat each campaign as a live learning system and carry lessons into the next one.

The findings add to a broader debate in advertising and marketing over whether AI will improve effectiveness or mainly increase production speed and volume. Here, the report suggests the difference will depend less on the tool itself than on whether marketers can move beyond static segmentation and use behavioural data from online communities to shape what they create.

Marcos Angelides, Managing Director, L'Oréal Lab & Head of AI Operations, Publicis Media, said: "Where the advantage comes is the data that you use to train those models, because AI is only as good as the data it's trained on. You've got to have behavioral data. You've got to know what people actually do, not just what they say they do."