AI citation study: Websites AI cites most in Home & Furniture

Prompts about the home range from the practical to the aesthetic, with a strong safety thread running through them.

• Choosing a retailer: "IKEA or Maisons du Monde for a sofa", "where to buy solid wood furniture that lasts".
• Solving a room problem: "how to reduce echo in an open-plan living room", "curtains that keep heat out in summer".
• Health and safety: "are scented candles bad for indoor air", "is an essential oil diffuser safe around a baby", "does a filter jug remove limescale".
• Gifts and style: "luxury candle for a housewarming", "Scandinavian decor ideas for a small flat".

The health questions set this industry apart. Products that stay in a room all day raise concerns about emissions, allergens or water quality, and AI models treat those concerns as topics that call for official sources.

Large home retailers such as IKEA, Leroy Merlin or Maisons du Monde are the default reference for furniture and DIY prompts. Their catalogues are vast, their product pages are detailed and their names already appear in countless articles, which makes them easy for a model to recommend.

Specialist brand sites (candles, diffusers, acoustic panels, water filtration) are cited when the prompt names a product type precisely. Health and environment agencies, for example ANSES and ADEME in France or the EPA in the United States, appear on indoor air, fragrance and water questions, together with standards bodies such as IFRA for fragrances or NSF for water filters.

Wikipedia, YouTube (room makeovers, assembly tutorials), Reddit and review platforms like Trustpilot supply background, demonstrations and buyer feedback. Comparison and test sites are cited for "best ... for ..." prompts.

When a prompt contains a health worry, a model tends to start with the precautions published by public agencies, then name products that fit those precautions. A candle brand that documents its wax, wicks and fragrance compliance, or a filter maker that publishes test results against a named standard, gives the model a reason to cite it right after the agency.

Ignoring the question is not a neutral choice. The answer will then be built from agency warnings and third-party tests alone, and the brand may be left out or mentioned with reservations.

Mass retailers win generic prompts almost by default. A smaller furniture maker or decor brand rarely beats them on "where to buy a sofa", but it can own precise requests: "made-to-measure acoustic panels", "solid oak bed made in France", "refillable diffuser without alcohol". The narrower the need, the more a detailed specialist page counts.

Inspiration platforms play a smaller role in citations than in discovery. AI answers quote text, so a beautiful catalogue still needs written descriptions, dimensions and materials before an assistant can use it.