A new Maverick Indonesia whitepaper has found that AI-generated answers about beauty and personal care in Indonesia draw on a much wider information ecosystem than brands’ own websites and channels.
Beauty in The Eyes of AI, produced in collaboration with Dream.co.id and Diadona, examined 8,000 AI-generated responses and 63,286 citations across ChatGPT, Google AI Overview, Perplexity, Copilot, and Gemini in August 2026.
"Beauty and personal care marketing needs to combine creativity with data. In this study, AI drew on sources well beyond beauty media, with different patterns depending on what consumers asked. These findings can sharpen how brands understand audiences and media credibility, which is the basis for choosing channels and developing messages," said Cahyanto Arie Wibowo, Director of Business Growth & Innovation at Maverick Indonesia.
The study tested 100 prompts across five generative AI platforms, including 80 general prompts and 20 prompts that mentioned a brand by name. Each prompt was run daily for 16 days, generating 8,000 responses in total.
All prompts were written in Indonesian and informed by search-volume patterns. The seven topics analysed were based on around 6.4 million Indonesian Google queries related to beauty and personal care, cross-checked against consumer purchase behaviour research.
The topics covered effectiveness and suitability, ingredients and formulation, reputation and reviews, availability and sales channels, price and value, safety and compliance, and beliefs and cultural fit.
For general beauty questions, AI drew more heavily on third-party sources than on brand-owned websites.
News Media and Health and Medical Information accounted for a combined 46.6 per cent of citations from general prompts, compared with 11.6 per cent from Brand Owned sources.
The report says news outlets and beauty publications provide broad, independent coverage of brands, products, and beauty topics, while health and medical platforms contribute information around ingredients, safety, and suitability.
Retail and marketplace sites were also significant sources, providing details such as pricing, product information, and availability.
Brand-owned websites remained part of the information mix, but the study found that they were only one part of the broader ecosystem used by AI to answer general beauty questions.
The source mix shifted when a specific brand was named in a prompt.
For branded questions, Retail and Marketplace and Brand Owned sources became more prominent, while News Media and Health and Medical Information no longer held the leading positions.
Social, Blog, and Community content also became more important. According to the report, this category rose from 9.2 per cent of citations in general prompts to 19.5 per cent when a brand was named.
The report says this reflects AI drawing more directly on product listings, brand websites, creator content, and community reviews when the question already points towards a specific brand or product.
The study found that AI does not limit its beauty-related sources to specialist beauty publications.
Suara.com, a general news publisher, received more News Media citations than Beautynesia, Female Daily, and Popbela combined.
The report says this shows that being a specialist beauty publisher does not necessarily determine whether a source will have the strongest presence in AI-generated answers.
For brands, the report recommends looking beyond traditional beauty media when building AI visibility, as the publishers shaping AI answers can come from a much broader media landscape.
Health and Medical Information was one of the strongest source categories in the study.
Halodoc appeared across 74 of the 80 general prompts, with content covering product recommendations, ingredients, safety, and practical skincare routines.
The report found that local health platforms were particularly prominent in general beauty questions, while international medical and clinical sources became more relevant when questions required stronger medical evidence, including around pregnancy, ingredient safety, effectiveness, or specific health conditions.
Maverick Indonesia says this creates an opportunity for brands to work with local health platforms on science-based content, while recognising that more specialised questions may draw on more specific medical sources.
The study also found that retail and marketplace sites are serving a role beyond product purchase.
Shopee, Zalora, and Watsons had the strongest presence among Retail and Marketplace sources cited in general beauty questions.
Many citations came from buying guides, comparisons, and recommendation content. The report says this means product information published through retail platforms can become part of the information AI uses when answering consumer questions.
In other words, the information attached to a product listing or retailer content can influence AI visibility as well as helpconsumers make a purchase.
One of the strongest findings is the difference between individual AI platforms.
News Media accounted for 53.1 per cent of Microsoft Copilot citations, compared with only 9.5 per cent of Google AI Overview citations.
Google AI Overview placed considerably more weight on Social, Blog, and Community content, which accounted for 28.1 per cent of citations, compared with 1.8 per cent for Copilot.
Perplexity and Gemini also leaned more heavily towards News Media, at 37.3 per cent and 33.5 per cent respectively.
ChatGPT showed a more distributed source mix. Health and Medical Information accounted for 23.7 per cent of its citations, Brand Owned sources at 19.5 per cent, News Media at18.3 per cent, and Regulatory, Government,and Certification sources at 15.1 per cent.
The report says these differences mean there is no single source strategy that will work in the same way across all AI platforms.
The study also found that source use changes according to the type of question being asked.
Effectiveness and Suitability questions drew 45.5 per cent of their citations from Health and Medical sources.
Price and Value questions drew 36.8 per cent from News Media, while Brand Owned sources reached their highest share, 21.3 per cent, in questions about Beliefs and Cultural Fit, including halal status, sustainability, and other values important to local consumers.
The report says this means brands cannot rely on a single source category to build AI visibility across every consumer question. Instead, content and source presence need to be aligned with the types of questions consumers are asking.
The research found that social sources cited by AI include more than traditional influencers.
Across the platforms studied, AI cited individual creators, doctors, and other professionals, including publishers, institutions, communities, and brand-owned social accounts.
Instagram and Lemon8 appeared across almost the full range of general beauty questions, while YouTube and TikTok were more prominent in questions about effectiveness and suitability. Reddit contributed consumer perspectives around beliefs and cultural fit.
On Instagram, four of the 10 most-cited profiles were individual creators. On YouTube, Female Daily was the most-cited channel, followed by an aesthetic doctor.
The report saysinfluencer activity is therefore only one part of social visibility and recommends that brands build presence across creators, professional sources, and their own content.
Content age also influenced the information cited by AI, but the study found that newer content was not the only source of visibility.
Around 61.3 per cent of citations came from pages published within the previous 12 months, while 38.7 per cent came from content more than a year old.
Around 17 per cent of citations came from pages more than three years old.
The report says this shows that established content can continue to contribute to AI visibility if it contains information relevant to the question being asked. It recommends maintaining both fresh and evergreen content rather than focusing on recency alone.
The report’s central finding is that AI visibility depends on how well a brand is represented across different information sources, not simply how much content the brand publishes itself.
For communications professionals, the report points to three priorities:
Keep core product and brand information clear on owned channels
Build credible information around the brand through news, medical, and other external sources
Maintain useful social, blog, and community content that captures consumer experiences and opinions.
It also cautions against simply producing more content across more channels. The report’s recommendation is to make sure information about the brand is clear, credible, relevant, and useful in the source types that AI is actually drawing from.