How the answers frame Unilever
This is a snapshot from 29 September 2026: 250 benchmark questions put to ChatGPT, Claude, Gemini, Perplexity and Grok, 1,250 answers in all. Unilever was named in 269 of them. Grok named it most often (84 of 250 answers) and Claude least (36 of 250). Among consumer packaged goods peers in this set, only Procter & Gamble appeared in more answers.
Visibility depended heavily on what the question asked. By question intent, Unilever appeared most in leadership (45%), employer (43%) and innovation (42%) questions. Sustainability (28%), sourcing transparency (25%) and packaging waste (24%) came next. It was nearly absent from value (0%), ingredient health (2%), service (5%) and reliability (6%) questions.
The framing was mixed. In the best-run question, ChatGPT placed Unilever second behind P&G and called it "a bit more uneven" historically. On career questions, Gemini pointed to its graduate leadership programme. Perplexity, citing an Oxfam scorecard, called Unilever a relative leader on supplier treatment. Claude's deforestation answer credited it with the top score among the brands it cited from Rainforest Action Network.
Unilever also appeared where the questions asked about problems. On the "avoid" intent it showed up in 35% of answers. Perplexity's greenwashing answer named its refill pouches. Gemini's answer on quality slipping after ownership changes used Breyers as its food example. Its ethics-ranking answer listed Unilever among "highly scrutinized" household product companies.
The answers' own credit tags were thin. The largest category was "Not Substantive", with seven tags, followed by Sustainability & Ethics with five.
A bit more uneven than P&G historically, but still one of the strongest global consumer-goods operators, with huge scale and broad international reach.
What the press is saying
PeakMetrics' news index logged 9,437 Unilever mentions in the last 30 days, 843 of them with the brand in the headline. That is mid-pack among the 15 peers in this set: Coca-Cola (22,781), Nestlé (12,010), PepsiCo (11,214) and General Mills (10,708) all have more.
The recent headlines point two ways. One is operational: Brazil halted cleaning product output at a Unilever plant. The other is a board appointment at Nuritas of a former Unilever North America president, syndicated across at least four outlets. In that story Unilever appears as a résumé line rather than the subject.
Oxfam’s comparative scorecard points to Unilever as a relative leader…
The gap
Press volume and AI visibility are different measures, and they diverge here. Coca-Cola has far more news coverage than Unilever yet appeared in only 151 of the 1,250 answers. Unilever has less coverage than Nestlé, PepsiCo and Coca-Cola but appeared in more answers than any of them except P&G.
The themes also diverge. The most recent headline about Unilever itself concerns a plant stoppage, an operations story. In these answers, reliability and service questions surfaced Unilever in single digits. The assistants' Unilever story ran through leadership, talent, innovation and sustainability, with packaging and ethics criticism attached. Nothing in the snapshot shows that the press item shaped any answer.
…criticized over refill pouches promoted as an eco-friendly option, which a campaign group said were difficult or impossible to recycle.
What to watch
A comms team could start with the question types where Unilever is strong and where it is nearly invisible. Reliability, value, service and ingredient-health questions barely surfaced it in this snapshot, which leaves those answers to peers and third-party sources.
The critical framings trace to named sources. Perplexity's greenwashing answer cited a campaign group on refill pouches, and its supplier answer relied on an Oxfam scorecard. Gemini's Breyers example rested on consumer commentary. Tracking which domains sit behind each framing shows where a response would need to land.
These are single runs of live answers and can shift between runs. Treat them as a baseline to re-measure, not a ranking.
