Unilever says AI makes 12 times its 2024 content at 90% lower cost

Unilever's marketing organisation now produces twelve times as much content as it did in 2024, at 90% lower cost, using what the company calls an AI-powered engine, Marketing Dive reported on 12 November 2026 from the consumer-goods group's investor event of 4 November. The report does not name the tool or give the spend in money.

The saving is not going to profit. Unilever said it has cut the cost per unit of content and improved its return on marketing investment by between 10% and 15% a year, and that the gain is not expected to reach its bottom line, according to Marketing Dive, which describes it as a cost of doing business on digital media platforms.

I believe improving our return on investment is fundamental to keeping our business model working, continuing to drive top-line growth and delivering modest margin expansion every year.

Fernando Fernandez, chief executive of Unilever, at the 4 November 2026 investor event, quoted by Marketing Dive

What Unilever reported

• Volume and cost. Twelve times the content of 2024 at 90% lower cost. The report does not say whether the 90% is per piece of content or in total; a later passage says Unilever reduced "the cost per unit of content".
• Shelf life. Video content now lasts four days on platforms such as Meta, Google, TikTok, Amazon and Walmart. After that its return is negative, not just smaller, Marketing Dive reports. That is why the volume has to rise.
• Creators. Unilever says it now works with 500,000 content creators, up from the 300,000 it reported before and from 20,000 before Fernandez joined, by Marketing Dive's account. It uses AI to find, brief, vet and measure them, and says their content returns 1.5 times what brand-made content does.
• AI assistants. To be recommended when shoppers ask a chatbot, Unilever tracks more than one million prompts (the questions people type into AI assistants) across more than 30 markets and is restructuring its product data so software can read it, with partners including Google. It says it improved in more than 90% of the brand and market positions it tracks over six months; the report does not define a position.
• Consumer research. Chief marketing officer Leandro Barreto said the company uses synthetic audiences, meaning AI-simulated consumer panels, to follow more than 60,000 consumer signals and refine ideas before real consumers see them. Unilever says concepts that took months to validate now take hours.

The people side

The report says nothing about what changed for Unilever's marketers or its agencies, how many people run the engine, or who approves AI-made content before it is published. It does describe the structure around them: Barreto becomes chief growth and marketing officer on 1 January, and a reorganised operating model with six global units follows in 2027. In February 2026 Marketing Dive reported a five-year Unilever deal with Google Cloud covering brand discovery, measurement and marketing, and an in-house design unit, Sketch Pro, that uses generative AI.

The catch

Every figure is Unilever's own, told to investors, with no method, no baseline in money and no tool cost. Purchase intent and brand recall rose "in some cases", with no number. Fernandez said the model does not yet cover every brand: "Have we done it in all our portfolio? No." Marketing Dive adds that being treated as an authority by AI shopping assistants is a gap where some competitors have outperformed Unilever.

What to do if you run marketing operations

• Before the next budget round, write the AI content case as cost per asset multiplied by assets needed per month. Unilever's unit cost fell and its volume rose twelvefold; the product of the two is the budget line.
• This month, take your last ten paid videos and find the day on each platform when return on spend turned negative. That number sets how much content you need, whatever it costs to make.
• If someone reports your brand's position in AI assistants, ask for the prompt list and the definition of a position before the figure goes to the board.