Amazon CEO says shopping agents like Meta’s Muse ‘not great’

Amazon CEO Andy Jassy said third-party shopping agents are not yet more convenient than buying directly from retailers, while noting they were not responsible for Amazon’s weaker-than-expected third-quarter advertising performance. He said more than 400 million customers used Amazon’s Alexa shopping agent in the past year, interactions grew 570% year over year, and U.S. users spent 50% more per order. Meta said more than 10 million people in the U.S. and Canada had set up its Muse assistant, with over a quarter of users shopping through it and average purchases topping $200.

Analysis

The economically important contest is ownership of shopping intent, not whether consumers use a chatbot. If an outside agent becomes the discovery layer, Amazon risks losing sponsored-product impressions and first-party signals even when the final order still ships through Amazon. That could pressure ad monetization before it materially changes retail volume. Conversely, an agent that sends incremental, high-conversion traffic may expand retailer sales; the merchant’s net benefit depends on customer-acquisition cost, attribution, and who retains customer data.

Amazon’s usage and higher-spend claims do not establish incremental profit: heavier shoppers may self-select into Alexa, and larger baskets do not show whether ad revenue or conversion improves. Meta’s setup count and shopping engagement likewise do not establish repeat usage, completed transactions, or monetization. Near term (days to weeks), this is more narrative than earnings evidence. Over 1–3 months, watch Amazon ad growth and sponsored placement economics alongside repeat use and merchant conversion for Muse. Over 6–18 months, the risk to Amazon rises if agents repeatedly route demand outside its owned interface; Amazon’s own assistant could instead defend the funnel if it drives incremental purchases without worsening economics.

No clean directional trade is supported yet. A conditional META-over-AMZN relative position is only justified by evidence of durable agent retention and merchant conversion alongside deterioration in Amazon advertising; it is falsified by weak repeat use, limited completed orders, or stabilization in Amazon ad growth. Retail partners may gain incremental demand, but could become more dependent on an intermediary and surrender customer insight. The key contrarian point: agent adoption is not equivalent to disintermediation, and blocking access may protect Amazon’s ad inventory while risking longer-run consumer preference.

Key Decisions for Investors

• Do not trade the usage figures alone. Track Amazon advertising growth and sponsored-product performance at the next report; a sustained deceleration would strengthen the disintermediation concern, while stabilization weakens it.
• Keep META-over-AMZN as a conditional relative-value watch, not an immediate position. Require evidence of Muse repeat use, completed purchases, and merchant conversion; reverse the thesis if engagement fails to translate into transactions or Amazon ad growth stabilizes.
• Monitor Shopify, Walmart, and Best Buy for evidence that agent-referred orders are incremental and economically attractive. Without conversion, repeat-purchase, and customer-acquisition data, treat partnership announcements as distribution tests rather than earnings catalysts.
• For Amazon, verify whether Alexa’s reported spend-per-order advantage reflects incremental orders and profitable ad placement rather than user mix. If not, the claimed defense of its shopping funnel is weaker than headline usage suggests.