Social Rocket
According to its own statements, Liquid Death has switched its retail media campaign work from weekly or twice-weekly reviews to a daily rhythm. The US beverage company uses an AI-supported data and decision platform with the agency Power Digital for this. Marketing lead Benoit Vatere reported this on a panel at Advertising Week New York on 5 October 2026. An independently measured sales increase from this change was not published.
The relevant development is therefore less another AI creative tool than a reorganisation of ongoing advertising operations: data analysis, recommended action, human approval and optimisation are meant to work more closely together.
The difference between a chatbot and campaign operations
Power Digital officially presented its technical architecture on 4 August 2026. According to a company announcement, it consists of three interconnected building blocks:
- Iris links business and performance data to identify growth obstacles and options for action.
- Omega turns recommendations into a shared workspace for the agency and clients; specialised agents support operational steps.
- Creative Affinity analyses creative material and is meant to better estimate its potential impact.
Power Digital describes a approval model with human decision-making before agents take action. This is essential: a system that automatically changes bids or budgets based on faulty signals can scale errors quickly. Greater speed is only useful if rights, data quality and decision boundaries are in place.
Liquid Death reports faster work – not proven ROI
According to Marketing Dive's report of 7 October, Vatere can now check campaign signals directly without first initiating manual queries or coordination. A second report on the panel also describes the shortened planning of Amazon activities. The statements come from the company and its agency environment.
Not proven from this alone are any specific cost savings, better advertising incrementality or additional sales. Likewise, Power Digital's claims about the size of its training data and above-average client growth are, for now, provider statements. A reliable proof of effectiveness would require before-and-after data with a suitable control group, comparable budgets and metrics such as incremental contribution margin.
The brand's apparent AI contradiction
Liquid Death has also implemented advertising ideas that satirically criticise the water and electricity demand of AI data centres. This does not mean the brand internally rejects generative AI. Public campaign messaging and operational technology decisions are different levels. What matters is whether a company can credibly justify both positions – and whether the promised savings are verifiable.
For marketing teams with limited staff, the transferable idea lies less in buying a large agency platform than in building a clean process: make platform and shop data centrally available, have unusual spikes flagged automatically, prioritise recommendations and approve sensitive budget changes. Our reports on AI agents with advertising budgets, retail media incrementality at Lowe's and AI content with human quality control show where the limits of a pure automation logic lie.
Which questions agencies need to answer
Before deploying comparable systems, advertisers should put in writing who may access customer data, which decisions are made automatically, how changes are logged, whether recommendations are based on traceable metrics, and which independent control measurements are planned. Without these answers, "AI-native" initially only increases technical complexity.
Sources: Power Digital, company announcement dated 04.08.2026; Marketing Dive, 07.10.2026; Marketers & Money, 07.10.2026.