Macy's AI chatbot users spend 4.75 times more than non-chatbot shoppers. That is the headline number from the first major retailer case study on AI-powered shopping assistance, and it should fundamentally change how every e-commerce brand thinks about AI's role in the purchase journey.
Ask Macy's launched on March 27, 2026, powered by Google Gemini. Within its first four days of operation, the data was already unambiguous: shoppers who engaged with the AI assistant converted at dramatically higher rates, bought more items per session, and explored more product categories than those who browsed traditionally. This is not a marginal improvement. It is a step change in e-commerce economics.
Here is what makes it work, what the data actually shows, and how brands of every size can apply these lessons.
What Is Ask Macy's and How Does It Work?
Ask Macy's is a conversational AI shopping assistant embedded directly into Macy's website and mobile app. It is powered by a fine-tuned version of Google Gemini trained on Macy's entire product catalog, decades of styling expertise, and customer interaction data.
The experience works like this: a shopper can type or speak a natural language query like "I need an outfit for a spring wedding in Napa, budget around $300" and receive personalized product recommendations with complete outfits, alternative options at different price points, and styling advice that draws on Macy's brand expertise.
Key capabilities include:
- Conversational product discovery: Natural language queries replace category navigation and filter menus
- Complete outfit building: The AI assembles full outfits across departments (clothing, shoes, accessories) rather than recommending individual items
- Image-based matching: Users can upload photos of items they like and receive visually similar options from Macy's inventory
- Occasion-based recommendations: The AI understands context (wedding, job interview, vacation) and adjusts recommendations accordingly
- Real-time inventory awareness: Recommendations only include items available in the user's size, preferred color, and delivery timeline
- Style memory: For logged-in users, the AI remembers past preferences and purchases to improve future recommendations
What Do the Spending Numbers Actually Show?
The 4.75x spending multiplier is the top-line figure, but the underlying data reveals more about why AI shopping assistance is so effective.
| Metric | Non-Chatbot Shoppers | Ask Macy's Users | Multiplier |
|---|---|---|---|
| Average order value | $67 | $142 | 2.1x |
| Items per order | 1.8 | 3.4 | 1.9x |
| Conversion rate | 2.3% | 8.7% | 3.8x |
| Session duration | 4.2 minutes | 11.8 minutes | 2.8x |
| Cross-category purchases | 12% of orders | 47% of orders | 3.9x |
| Return rate | 24% | 16% | 0.67x (33% lower) |
| Total spending per visitor | $1.54 | $7.32 | 4.75x |
The 4.75x total spending lift is a function of both higher conversion rates and larger basket sizes. But the most revealing metric is the cross-category purchase rate. Nearly half of AI chatbot orders include items from multiple departments, compared to just 12% of standard orders. The AI is doing what the best in-store sales associates do: building complete solutions rather than selling individual items.
The lower return rate is equally significant. AI-assisted purchases result in 33% fewer returns, likely because the conversational process helps shoppers make better-informed decisions and avoid impulse purchases that do not fit their actual needs.
Why Does AI Shopping Assistance Drive So Much More Spending?
The 4.75x multiplier is not magic. It is the result of AI solving four specific problems that have plagued e-commerce since its inception.
The Navigation Problem
Traditional e-commerce forces shoppers to navigate hierarchical category trees and manipulate multiple filters to find products. This works for shoppers who know exactly what they want but fails for the majority who have a general need but not a specific product in mind.
Research shows that 68% of e-commerce sessions start without a specific product in mind. These browse-intent sessions have historically had conversion rates below 1%. AI shopping assistance converts these browse sessions into guided discovery experiences that feel more like working with a knowledgeable personal shopper.
The Context Problem
A filter menu cannot understand "I need something for my daughter's college graduation" the way a human shopping assistant can. Traditional e-commerce strips away the context that drives purchase decisions: the occasion, the personal style, the budget constraints, the practical requirements.
Ask Macy's processes all of this context simultaneously. A single conversational exchange conveys more intent information than 15 filter selections. This contextual understanding leads directly to more relevant recommendations and higher conversion rates.
The Cross-Sell Problem
E-commerce sites have always struggled with cross-selling. "Customers also bought" widgets generate modest incremental revenue, but they cannot build a coherent outfit or design a complete room the way a skilled associate can.
AI shopping assistants solve this by understanding how products work together. When the AI recommends a dress, it simultaneously suggests shoes, a bag, and jewelry that complement the selection. This is not random cross-selling. It is contextual outfit building that adds genuine value to the shopping experience.
The Confidence Problem
One of the biggest barriers to e-commerce conversion is purchase uncertainty. Shoppers are not sure the item will look right, fit correctly, or match their other belongings. They add items to cart and then abandon them.
AI shopping assistants reduce uncertainty by providing personalized styling advice, explaining why specific items were recommended, and offering alternatives if a shopper expresses doubt. This guidance function builds the confidence that converts browsers into buyers.
What Are the Category-Specific Results?
Macy's has shared category-level data that reveals where AI shopping assistance has the most impact.
| Category | Spending Multiplier | Conversion Lift | Avg. Basket Increase |
|---|---|---|---|
| Women's fashion | 5.2x | 4.1x | 2.3x |
| Home furnishing | 4.9x | 3.6x | 2.8x |
| Men's fashion | 4.6x | 3.5x | 2.0x |
| Accessories | 3.8x | 2.9x | 1.7x |
| Beauty | 3.6x | 3.2x | 1.4x |
The strongest results appear in categories with high complexity and high consideration: fashion and home furnishing. These are categories where shoppers benefit most from expert guidance and where cross-selling opportunities are richest.
Beauty shows a lower spending multiplier but a proportionally strong conversion lift, suggesting that AI assistance is particularly effective at converting beauty browsers into buyers even if the average order value increase is modest.
How Can E-Commerce Brands Replicate These Results?
You do not need Macy's budget to implement AI shopping assistance. The core principles that drive the 4.75x spending lift are applicable to brands of any size.
Step 1: Get Your Product Data AI-Ready
AI shopping assistants are only as good as the product data they draw from. Before implementing any AI chat tool, audit your product data for:
- Complete attribute coverage: Every product needs detailed specifications, materials, dimensions, and use cases documented in structured fields
- Occasion and context tags: Tag products with the occasions and contexts they serve (wedding, casual office, outdoor summer, etc.)
- Compatibility data: Document which products work together (matching items, complementary products, required accessories)
- Honest limitations: Include sizing notes, care requirements, and situations where a product may not be the best choice. This builds the trust that drives conversion.
Step 2: Choose the Right AI Platform
Several platforms now offer AI shopping assistant capabilities for e-commerce brands:
| Platform | Best For | Price Range | Key Capability |
|---|---|---|---|
| Shopify Sidekick | Shopify stores | Included with Plus | Native Shopify integration |
| Google Cloud Retail AI | Mid-market to enterprise | $2,000-$15,000/mo | Gemini-powered, same tech as Ask Macy's |
| Amazon Personalize | Amazon sellers | Usage-based | Deep e-commerce ML models |
| Custom Gemini/GPT build | Large brands | $10,000-$50,000+ setup | Full customization |
| Tidio AI | Small to mid-market | $29-$394/mo | Quick deployment, good starter option |
Step 3: Train the AI on Your Brand Expertise
The Ask Macy's chatbot is not just a generic product recommender. It is trained on Macy's specific styling expertise and brand knowledge. Your AI assistant should reflect your brand's unique expertise:
- Upload your buying guides, style guides, and expert content as training material
- Define your brand's recommendation philosophy (do you emphasize value, luxury, sustainability, performance?)
- Create scenario-based training examples that reflect how your best human sales associates handle common customer questions
- Establish guardrails for what the AI should and should not recommend
Step 4: Optimize for Cross-Category Discovery
The 3.9x increase in cross-category purchases is a massive revenue driver. Configure your AI assistant to:
- Proactively suggest complementary products from other categories
- Build complete solutions rather than recommending individual items
- Ask context questions that open up cross-category opportunities ("What occasion is this for?" naturally leads to outfit building)
- Present cross-category bundles with clear value explanations
Step 5: Measure and Iterate
Track these specific metrics from day one:
- AI engagement rate: What percentage of visitors interact with the chatbot?
- Conversion rate lift: Compare chatbot-engaged visitors to non-engaged visitors
- Average order value change: Are chatbot users spending more per order?
- Cross-sell rate: Are chatbot sessions generating multi-category purchases?
- Return rate differential: Are AI-assisted purchases returned less frequently?
- Session duration: Longer is better if it correlates with higher conversion
How Does This Connect to the Broader AI Shopping Trend?
Macy's is the first major case study, but it validates patterns we see across the entire AI shopping landscape.
Amazon Rufus reaches 250 million users and drives an estimated $10 billion in incremental sales by providing AI-powered product discovery within the world's largest e-commerce ecosystem. Perplexity processes over 100 million monthly queries, with an increasing share related to product research and recommendations. ChatGPT's 900 million weekly users generate millions of product discovery conversations daily.
The through-line across all of these platforms: AI-assisted shopping drives dramatically better outcomes than self-service browsing. The 60% conversion improvement observed across AI assistant engagements aligns with Macy's data. Consumers who receive AI-powered guidance buy more, buy better, and return less.
AI shopping is projected to influence $20.9 billion in commerce in 2026, a 4x increase year over year. The brands that implement AI shopping experiences now, whether through their own chatbots, optimization for external AI platforms, or both, will capture a disproportionate share of that growth.
The 4.75x spending multiplier is not a Macy's-specific phenomenon. It is a preview of what AI-assisted commerce looks like at scale.
Want to explore how AI shopping assistance could impact your e-commerce metrics? Schedule a free consultation with our team. We will analyze your product catalog, customer journey, and competitive landscape to estimate the potential revenue impact of AI shopping features for your specific business.