The strategy
PUMA used preezie’s AI Shopping Assistant to support customers during shopping moments across the website.
The strategy focused on helping shoppers:
- Get quick answers to sizing and fit questions
- Compare running technologies such as comfort vs. performance
- Understand materials and technical specs
- Get immediate answers to simple customer service questions
- Discover the right product faster
- Reduce hesitation before purchase
It also gave PUMA an operational advantage: a solution that could work with minimal manual effort while still delivering relevant, page-aware responses.
The launch & early insights
PUMA’s AI Shopping Assistant went live in 8 weeks and quickly became an active part of the customer journey. During the reporting period, more than 30,000 unique users engaged with AI, sending 64,000+ messages.
The insights were especially clear:
- 42% of customer questions were about sizing and fit
- 20% focused on cushioning and comfort
- 19% were about fabric and materials
- 10% related to colour and style
- 9% covered technical specs
This showed that customers were not just browsing. They were asking the kinds of questions that typically block a purchase / customers need answered to feel confident buying.
Business impact
The commercial impact from AI was clear across every level of intent.
Across all users:
AI-assisted shoppers converted at 7x the rate of shoppers who did not use AI.
Furthermore revenue per visitor was 8x higher for AI users overall.
Even in higher-intent segments, AI continued to lift performance. For example when we compare those who searched vs those who used the AI, conversion rate was still 3 times higher!
What PUMA customers were asking
The AI data surfaced clear customer needs and practical opportunities for both eCommerce and customer service.
The biggest friction point was sizing. Customers were regularly asking things like:
- Are these true to size?
- What are the dimensions?
- How do AU/US/UK/EU sizes compare?
Other common pre-purchase questions focused on:
- Material and fabric details
- Leather vs mesh decisions
- Cushioning and comfort
- Arch support and performance features
- Product weight
- Wide fit availability
For PUMA, that meant the assistant was doing two valuable jobs at once: helping shoppers get answers immediately, and reducing the flow of simple questions that might otherwise end up with the customer service team.
Beyond performance: insight for the wider business
One of the strongest additional benefits for PUMA has been the value of transcript insights.
By reviewing customer conversations, the team has been able to extract useful themes and share them across the business. That includes:
- What customers are asking most often
- Which products come up most
- Where product information may be missing or unclear
- Recurring questions outside core product discovery, including in areas like promotions, returns and shipping
This gives cross-functional teams a clearer view of real customer intent and friction points, using actual shopper language rather than assumptions.
Product and merchandising insights
- AI Search behaviour also showed strong demand in running, with 8 of the top 10 searches tied to running products. Customers were actively comparing running technologies, indicating strong interest in comfort-versus-performance decisions
That gives PUMA a clear roadmap for:
- Merchandising
- Content improvement
- Product education
- Future category expansion
Summary
- The AI Shopping assistant provides quick accurate answers to customers, allowing them to benefit from page aware, immediate, always-current responses.
PUMA is benefiting from page-aware, immediate, always-current responses — made the solution especially valuable.
With preezie AI, PUMA:
- Increased conversion by 7x
- Improved AOV by 6.4%
- Maintained a 99% response success rate
- - Saw conversions 3 times higher then search
- Reduced friction from simple service and product queries
- Surfaced valuable transcript insights for cross-functional teams
- Identified merchandising and category opportunities
