Rufus, Amazon's AI shopping assistant, is now answering a non-trivial share of category-level shopper questions. For brands, that means a new surface where you are either the answer or you are not in the answer at all.
We sampled 200 category prompts across CPG, electronics, kitchen and pet over six weeks. We logged which brands Rufus surfaced and what evidence it cited.
What Rufus actually cites
Across 200 prompts, Rufus's citations came from, in order of frequency:
- Product titles — heavily weighted. Long, attribute-rich titles win.
- Bullet points — second most cited. Concrete claims beat marketing language.
- Review text — Rufus surfaces direct phrases from reviews. Sparse review pools rarely cite.
- A+ content text — surprising: image-heavy A+ modules with thin alt-text are not cited.
- Q&A section — long-tail prompts pull from Q&A more than from product description.
What it does not cite (much)
- Brand homepage copy on the storefront tab
- Lifestyle imagery without text
- Generic marketing language ("premium quality", "best in class")
The implication for brand operations
If your listing strategy is built around brand storytelling — beautifully shot A+ content, narrative bullets, minimal Q&A — Rufus will skip you. The brands Rufus cites have:
- Attribute-dense titles that read almost like specifications.
- Bullets that make claims with units (e.g. "pH-balanced 5.5", "120 wash cycles").
- Active Q&A pools — 30+ questions, brand-answered.
- Review volume above category median, recently active.
What to do this month
- Re-write top-SKU titles to lead with attributes, not brand voice.
- Audit bullet copy: replace adjectives with measurable claims.
- Seed Q&A — brands can answer their own questions; do it.
- Make sure A+ content has substantive alt-text.
Rufus is the new search query parser. Optimize for it the way you optimized for search keywords ten years ago — except now the surface is conversational.
Supersonic Supply runs Rufus-visibility audits as part of every Brand Audit. Request one.
