Has your business started decoding Rufus yet?
Mars United Commerce and Profitero+ are partnering with our clients to undertake an exciting research project designed to identify the patterns behind the products being recommended by Amazon’s Rufus agentic AI search engine.
Our initial analysis of over 1,000 products recommended from roughly 300 prompts is already providing insights into the role that customer reviews play in the recommendation process:
- Rufus only recommends products with a rating of 4 stars or higher.
- The average number of reviews for recommended products is roughly 9,000.
- Products with low review counts are rarely recommended (items with only 1 review only 0.2% of the time).
But that’s just the beginning. We’ll share more findings beyond customer reviews as we move forward with the research. For now, here’s the big question: What else is moving the needle to influence Rufus’ recommendations?
Our ultimate goal is to help clients move from observation, to insights, to action through the unique combination of powerful insights from Profitero+ and activation services from Mars United’s world-class experts.
Rankings vs. Relevance
Elsewhere, when we compared Rufus results with Amazon.com’s page-1 search results for the exact same prompts over a two-month period, we uncovered two powerful findings:
1. Only 22% of products on page 1 of search also appeared in Rufus results. Even high-ranking products aren’t guaranteed visibility. Rufus is extremely selective, surfacing only a small fraction of what Amazon.com’s algorithm ranks highest.
2. 36% of products recommended by Rufus weren’t even on page 1. Rufus isn’t just echoing the Amazon search engine. It’s pulling in products beyond page-1 visibility, which suggests that it’s optimizing for perceived relevance, not just rank.
This data may be good news for challenger brands trying to break through, but potentially bad news for category incumbents.
But in these still early days of on-site agents, we don’t yet know how well Rufus’ recommendations drive conversion. For now, the technology sits somewhere between guide and gatekeeper — helping to shape what shoppers consider without fully replacing search.
As the popularity of AI assistants grows, however, it’s clear that getting on page 1 of search results is still necessary — but no longer enough.
The goal, therefore, is to build on SEO:
- Audit and maintain your online product catalog elements to ensure retail readiness.
- Keep optimizing PDP content for search engine discoverability.
- Make sure your text-based PDP content is clear, conversational, and touches on the most important shopper needs and questions.
To be among the first to get the full research report later this year, you can sign up here to receive communication from Profitero+: https://www.profitero.com/resources.
