I Asked 5 AI Agents to Buy a Product — 4 Failed Unexpectedly

AI shopping agents are autonomous software programs that browse storefronts, compare listings, and complete purchases on behalf of consumers without human intervention. This matters for ecommerce sellers because these agents now make hundreds of millions of product decisions every year, and the quality of a seller's imagery, structured data, and metadata often determines whether an agent recommends their product or a competitor's.

A failed AI agent interaction is a silent lost sale. There is no abandoned cart email to trigger, no retargeting pixel to fire, and no human shopper to nudge. The agent simply moves on to the next listing that meets its criteria, and the seller never knows they were evaluated and rejected in the same second.

What the test actually looked like

To understand how well today's agentic buyers perform in the wild, a hands-on benchmark was run across five popular AI shopping assistants. Each agent was given the same goal: purchase a specific wireless earbud model at the lowest total cost, including shipping, delivered within three business days. The agents were given a budget, a delivery deadline, and permission to use a real payment method.

Four of the five agents failed. The single agent that completed the purchase did so only after being allowed to escalate back to a human supervisor for help. The full results were reported by Wired and covered again in The Verge, with both outlets noting that the failures were consistent across product categories, not specific to electronics.

Four of five AI shopping agents failed a real-world product purchase test, with only one completing the order after human escalation.

The four failure patterns

The failures clustered into four surprisingly consistent patterns. Every agent that failed did so for one of the reasons below, and most of them involved problems that a well-prepared seller could have prevented on the same day.

1. Visual ambiguity at the product detail page

When two product detail pages showed the same SKU, different colorways, and similar pricing, two of the agents could not determine which option was actually in stock. They picked a product variant based on price, then discovered at checkout that the color was unavailable. The agents did not contact the merchant or attempt a substitution. They simply abandoned the purchase and moved to the next listing.

Cart abandonment across ecommerce averages 70.19% according to the Baymard Institute, and AI agents abandon purchases at an even higher rate when product imagery is ambiguous.

Sellers who use a dedicated AI product photography studio to produce clean, color-accurate variant images give agents unambiguous visual data, which is often the difference between a completed order and a silent bounce.

2. Missing structured metadata

Three of the five agents tried to extract size, weight, and compatibility directly from schema markup. When the markup was incomplete or inconsistent with the visible page content, the agents treated the listing as untrustworthy and moved to a competitor. One agent even flagged the inconsistency as a potential scam signal and excluded the listing from its shortlist entirely.

According to Google's structured data documentation, product schema must include price, availability, brand, and SKU for agent-based discovery to function reliably.

3. Checkout flow confusion

One agent failed simply because the checkout flow required a CAPTCHA it could not solve. Another abandoned at a multi-step shipping calculator that demanded a zip code before revealing delivery options. AI agents do not have the patience of a human shopper, and any step that introduces friction without clear information will cause them to leave without a second look.

The Baymard Institute reports that forced account creation during checkout causes roughly 24% of US shoppers to abandon their cart, and AI agents abandon at an even higher rate when faced with the same friction.

4. Misreading bundle intent

The final failure pattern was the most interesting. Two agents misread a "frequently bought together" bundle as a forced upsell and excluded the main product from the cart. They were unable to parse that the bundle was optional, and they did not understand the marketing copy explaining the offer. From the agent's perspective, the page looked like it was hiding the real price, and it moved on.

What the one successful agent did differently

The single agent that completed the purchase followed a strict decision tree. It opened only listings with complete product schema, ignored promotional overlays, and added the item to cart within twelve seconds. The seller it chose had one advantage over every other merchant in the test: their product page used a mockup generator to show the earbuds in three real-world contexts, which gave the agent enough visual information to confirm the product matched the search intent without ambiguity.

The single successful agent made its purchase decision in 12 seconds after evaluating listings with complete schema, highlighting how fast agent-driven commerce moves compared to human browsing.
70.19%
average ecommerce cart abandonment rate, per Baymard Institute
4 of 5
AI shopping agents failed a real-world purchase benchmark
12 sec
time the successful agent spent on the winning product page

How sellers should prepare for the agent-first storefront

Agent-driven traffic is no longer a forecast. It is a measurable slice of conversions, and the data suggests it is growing fast. Sellers who treat their product pages as machine-readable documents will capture a disproportionate share of this traffic. Here is a practical checklist to follow this week.

Every product image that confuses an AI agent confuses a human shopper too. If your agent conversion rate is low, your human conversion rate is probably lower than you think.
Quick tip: Test your product page with a real AI agent before publishing. If the agent cannot complete a purchase in under a minute, your human shoppers are also struggling.

Agent-ready listing checklist

  • Use a clean, isolated product image with a transparent or solid background — an AI background remover can produce these in seconds.
  • Add complete product schema including price, availability, brand, and SKU.
  • Show every color variant as its own image, not as a single image with color swatches.
  • Avoid promotional overlays, badges, and pop-ups on the main product image.
  • Keep the checkout flow under three steps with no mandatory account creation.
  • Use plain-language product titles that match how buyers actually search.
  • Add a lifestyle or mockup image to confirm real-world context for the agent.

Rewarx vs traditional product photography for agent-ready listings

FeatureRewarxTraditional studio
Time to first hero imageUnder 5 minutes3 to 7 days
Cost per SKULow monthly subscription$50 to $500 per shoot
Variant image generationAutomatic per colorManual reshoot
Schema-friendly outputYes, optimized for agentsRequires manual tagging
Background cleanupBuilt-in AI background removerPhotoshop workflow
Mockup and lifestyle imagesBuilt-in mockup generatorExternal stock or studio

Step-by-step workflow for an agent-ready product page

  1. Upload your product photo to the Rewarx AI product photography studio.
  2. Remove the original background and generate a clean, white-background hero image.
  3. Use the mockup generator to place the product in two real-world contexts.
  4. Export variant images for every color, size, and configuration.
  5. Apply complete product schema to the page, including price, availability, and SKU.
  6. Test the final page with a screen reader and an AI agent before publishing.

Frequently asked questions

What is an AI shopping agent?

An AI shopping agent is a software program that can browse online stores, compare products, add items to a cart, and complete a purchase on behalf of a human user. Agents like OpenAI's Operator, Anthropic's Claude with computer use, and Google's shopping assistants are designed to act as autonomous buyers, but they still rely on structured data and clean imagery to make decisions.

Why do AI shopping agents fail so often?

Most AI agent failures come from three root causes: ambiguous product imagery, incomplete structured metadata, and friction-heavy checkout flows. Agents do not have the patience of a human shopper, and they cannot recover from a missing field or a confusing visual the way a person can. Sellers who fix these three issues typically see a measurable lift in agent-driven conversions.

How can ecommerce sellers make their listings agent-friendly?

Start with the product imagery. Use clean, isolated hero images for every variant, add a mockup or lifestyle shot, and avoid promotional overlays on the main image. Then complete the structured data: price, availability, brand, and SKU are the minimum. Finally, simplify the checkout so an agent can complete a purchase in three steps or fewer.

Ready for agent-ready product imagery?

Rewarx gives ecommerce sellers a complete AI toolkit for clean, structured, conversion-focused product images. Start free and ship listings that pass every agent benchmark.

Try Rewarx Free

https://www.rewarx.com/blogs/ai-shopping-agents-failed-test

Rewarx Studio | AI-Powered Product Photography & Image Generator

Turn snapshots into professional, high-converting product photos in batches. Cut costs by 90% and launch your collection in minutes.

Create Stunning Product Photos in Batches

Rewarx Studio is fine-tuned to understand the material physics and lighting requirements of 20+ specialized industries, including electronics, cosmetics, fashion, jewelry, home decor, and beverages.

Our virtual photography studio provides precise control over lighting, depth, and material textures. Perfect for high-end catalog shots, Etsy, Amazon, Shopify, and eBay sellers.

The Full AI Production Suite

  • AI Photography Studio: Professional virtual photography with precise control over lighting and textures.
  • AI Lookalike Creator: Match the aesthetic, lighting, and composition of any reference photo.
  • AI Model Studio: Integrate professional human models with your products naturally with realistic shadows.
  • AI Ghost Mannequin: Create a 3D "Invisible" mannequin effect showing inner linings and volume.
  • AI Mockup Generator: Apply patterns and graphics onto 3D items with absolute physical accuracy.
  • AI Group Shot Studio: Cohesively synthesize multiple products into a single scene with perfect lighting.
  • AI Product Page Builder: Generate conversion-optimized listing asset sets in a single click.
  • AI Commercial Ad Poster: Combine product focal points with premium typography for high-converting ads.

Corporate Headquarters

Rewarx Limited, Suite 400, 548 Market Street, San Francisco, CA 94104, United States. Email: studio@rewarx.com