Introduction
Shopper Intelligence gives you a comprehensive view of how customers interact with your AI Concierge. By analyzing customer conversations, emotions, and behaviors, this dashboard helps you identify opportunities to improve product discovery, enhance support, and increase conversions.
What You'll Find in Shopper Intelligence
Forms of Assistance
What it shows: A breakdown of topics your AI Concierge is handling in customer conversations, with intelligent insights into what your customers need help with most.
Understanding the Data:
The Forms of Assistance table displays conversation topics with total inquiry counts prominently shown for each topic. This gives you immediate visibility into how many conversations relate to each subject area, helping you quickly identify high-volume topics that matter most to your customers.
How to use it:
Review total inquiry counts - Each topic shows the total number of conversations related to that subject, helping you prioritize which areas to focus on
Check the "% of conversations" column - Understand what proportion of your customer interactions relate to each topic
Review "Reported as helpful/unhelpful" metrics - Identify areas where your AI is performing well or needs improvement
Use the tabs to filter between different conversation contexts:
All: Shows all conversation topics regardless of context
Unhelpful: Filters to show only topics marked as unhelpful by customers
Sales: Shows topics related to product questions, recommendations, and purchasing
Support: Shows topics related to customer service issues, returns, and assistance
Understanding Overlapping Data:
It's important to understand that topics can overlap - this is by design and helps provide complete insights:
One conversation, multiple topics: A single customer conversation may address several subjects. For example, a shopper asking "Do you have this in blue? What's your return policy?" would be tagged with both "Product Availability" and "Returns/Exchanges"
Total counts may exceed conversation totals: Because of this overlap, when you add up all topic counts, the sum may be higher than your total conversation count. This is normal and expected
"Unhelpful" metrics context: When reviewing "unhelpful" ratings, remember that some of these may come from the same conversations. For example, if a conversation covers 3 topics and receives an unhelpful rating, all 3 topics will show that unhelpful feedback
Pro Tips:
Click on any topic row to navigate to the Conversations page with that topic pre-filtered, allowing you to review specific conversation examples and understand the context behind the numbers
Focus on high-volume topics first - These represent the most common customer needs and offer the biggest opportunity for optimization
Compare inquiry volume to unhelpful ratings - A topic with 500 inquiries and 10 unhelpful ratings (2%) is performing much better than it might initially appear. Always consider the ratio, not just the absolute number
Look for patterns over time - Track how topics evolve as your business changes, seasons shift, or you launch new products
This section helps you understand what customers are asking about most frequently, identify emerging trends in customer needs, and measure how effectively your AI is resolving different types of inquiries.
Visitor Drop-Off Reasons (Beta)
What it shows: AI-generated categorization explaining why shoppers didn't complete their purchase.
How to use it:
Review the most common drop-off reasons in the chart
Check the conversation count and percentage for each issue
Focus on high-percentage issues first to make the biggest impact
Common drop-off reasons might include product availability issues, complicated exchange processes, or lack of product information. Addressing these issues can help reduce abandoned carts and increase conversions.
Shopper Emotion Analysis (Beta)
What it shows: The emotional motivation or tone detected in customer conversations when they initially engage with your AI.
How to use it:
See what emotions drive customers to initially engage with your AI
Identify emotional patterns that might affect purchasing decisions
Use this insight to adjust your AI's tone and approach
This analysis reveals the initial sentiment of shoppers when they begin a conversation - before any interaction with your AI Concierge. Understanding these starting emotions like "Seeking Clarity," "Needs Assistance," or "High Interest" helps you identify common customer mindsets and optimize your experience accordingly.
By analyzing these initial emotions, you can better understand what brings customers to seek assistance and tailor your AI's responses to address these starting points effectively.
Top 10 Products Added to Cart by the Concierge
What it shows: Products most frequently added to carts through AI-assisted conversations.
How to use it:
Identify which products convert best through AI conversations
Compare this data with your overall best-sellers
Consider featuring high-converting products more prominently
This insight helps you understand which products perform best when recommended by your AI Concierge.
Top 10 PDP Redirects by the Concierge
What it shows: Product pages that customers most frequently visit after AI recommendations.
How to use it:
See which products generate high customer interest
Identify products that customers want to learn more about
Compare with cart additions to spot conversion opportunities
This section highlights products that generate interest but may need additional information or incentives to convert to sales.
Making the Most of Shopper Intelligence
For best results, follow this workflow:
Start with "Forms of Assistance" to understand what customers need help with most
Check "Drop-Off Reasons" to identify and fix major conversion blockers
Use "Emotion Analysis" to evaluate customer initial sentiments and adjust your approach
Review top products to optimize your catalog and AI training
Troubleshooting
Issue: Data inconsistencies between charts
Ensure you're using the same date range for all comparisons
Remember that some metrics only track web conversations (not email)
Issue: Percentages don't add up to exactly 100%
This is normal due to rounding in the display and won't affect your analysis
Issue: Some conversations appear without topics
This may happen for very short interactions or when topics can't be confidently determined
Issue: Total topic numbers seem higher than total conversations
This is expected because one conversation can have multiple topics assigned to it