Unilever\u2019s been building a generative AI assembly line for its digital creative. The system might offer a model for heavyweight brand peers, but could also cause a headache for creative agencies.<\/p>\n
For the last year the CPG giant has been working with marketing services firm The Brandtech Group to build up its Beauty AI Studio: a bespoke, in-house system inside its beauty and wellbeing business. Now in place across 18 different markets (the U.S. and U.K. among them), the studio is being used to make assets for paid social, programmatic display inventory and e-commerce usage across brands including Dove Intensive Repair, TRESemme Lamellar Shine and Vaseline Gluta Hya.<\/p>\n
\u201cIt\u2019s a different way of working. We used to send briefs off and get content back. Now it\u2019s this agile, iterative approach,\u201d Selina Sykes, global vp and head of marketing transformation, beauty and wellbeing, told Digiday. Sykes estimated that on average, Unilever was using the system to create 400 creative assets per product: \u201cBefore, we\u2019d be doing 20 assets per campaign, and now we\u2019re doing hundreds.\u201d<\/p>\n
What\u2019s Unilever actually done?\u00a0<\/p>\n
The system relies on Pencil Pro<\/a>, a generative AI application developed by Brandtech Group. The tool draws on several large language models (LLMs), as well as API access to Meta and TikTok for effectiveness measurement. It\u2019s already used by hearing-care brand Amplifon to rapidly produce text and image assets for digital ad channels.\u00a0<\/p>\n In Unilever\u2019s process, marketers use prompts and their own insights about target audiences to generate images and video based on 3D renders of each product, a practice sometimes referred to as \u201cdigital twinning.\u201d Each brand in a given market is assigned a \u201cBrandDNAi\u201d \u2014 an AI tool that can retrieve information about brand guidelines and relevant regulations and that provides further limitations to the generative process. (For the tech heads, it does this using a cloud-based platform that combines LLMs, RAG frameworks<\/a> and REST, or representational state transfer, APIs.)<\/p>\n Sykes said the system would not be used to create any images of people: \u201cWe\u2019re not currently doing any human AI generation. We\u2019re doing it all around the product.\u201d She added that the company had used the system to produce creative assets 30% faster than its previous way of working, while other measures of advertising effectiveness such as video completion rate (VCR) and click-through rate (CTR) had doubled.\u00a0<\/p>\n Sykes didn\u2019t provide an estimate of the cash savings associated with the production efficiencies, but Unilever will be aiming for double-digit percentage savings on budgets. Per Medialink<\/a>, advertisers and agencies that have adopted AI into their processes have cut their creative production spending by 27%. And given that Unilever is one of the planet\u2019s largest advertisers spending \u00a37.8 billion ($10.42 billion) on marketing last year, per its 2024 financial results, even small figures imply significant cost reductions.<\/p>\n AI Beauty Studio hasn\u2019t just been an investment in tech for Unilever. It\u2019s meant committing to the time required to train and organize teams, too. The company first began working with Brandtech back in 2023, and began testing Pencil in first quarter 2024; much of the last year was spent designing and then administering Beauty AI Studio alongside in-housing specialist Oliver<\/a>. And 25,000 staff have been given \u201cAI fluency training\u201d so far.<\/p>\n So, what does this mean for creative agencies?<\/p>\n At one point, Unilever had roughly 3,000 agencies on its global roster. But it\u2019s spent much of the last decade<\/a> working to reduce that figure in 2024, when it consolidated much of its beauty and wellbeing business with British holding company WPP<\/a>. (Its media business remains scattered across a wide swath of shops, however<\/a>).<\/p>\n Gartner analyst Nicole Greene said that advertisers are likely to opt for \u201chybrid\u201d approaches to AI implementation, using agencies as vanguards of AI rather than jumping into the breach themselves. Major agency groups like WPP and Dentsu have each made AI-enabled production shops key parts of their offerings, emphasizing the capabilities of their respective Hogarth<\/a> and Tag<\/a> units.<\/p>\n Still, guidelines around responsible use of the tech, transparency with consumers over AI usage, data siloes and an uncertain regulatory environment<\/a> are all holding advertisers back from further implementation, Green said. Despite a 2024 Gartner study found 30% of marketing leaders planned to shift more marketing work in-house over the next 12 months, another study by the consultancy suggested that only 62% of marketing staffers have had training using generative AI.<\/p>\n For Forrester principal analyst Jay Pattisall, those stumbling blocks might actually \u201cdecelerate\u201d brands hoping to in-house. \u201cYou\u2019ll need the agency to implement, you\u2019ll need the agency to run, you\u2019ll need the agency to transfer and train, and that takes time,\u201d he said. Several CPG clients, including General Motors, Mondelez and Coca-Cola have opted thus far to outsource AI production capabilities to their agency partners (Monks, Publicis, Accenture and WPP for the record) rather than drag them indoors, Pattisall noted.<\/p>\n