The Headline
28,642 attendees engaged with Event Copilot AI, generating 64,119 interactions with 89% quick resolution — demonstrating strong adoption of conversational support during the event.
Users Served
28,642
Unique attendees
Interactions
64,119
Total questions asked
Quick Resolution
89%
Resolved in 1-2 messages
Est. Staff Deflection
12,824+
20-30% would have asked staff
Estimated breakdown of how 64,119 queries would have been handled without Event Copilot
20-30%
Would Ask Staff
12,824 - 19,236 queries
High-value deflections representing real workload reduction
40-50%
Would Self-Search
25,648 - 32,060 queries
Website, Google, app - but with fragmented results
20-30%
Might Not Ask
12,824 - 19,236 queries
Would give up or figure out through trial/error
Peak support: 5-7 PM
"Badge Pickup" was the most common question type
"Where is the photos showing"
"When is the next show"
"Transportation"
Pre-Event
Planning
Schedule & Registration
During Event
Navigation
Real-time Assistance
Post-Event
Follow-up
Resources & Next Steps
Recommendation for CES 2027
Better signage, earlier communication about pickup locations
Methodology & Research Sources
Industry research indicates that while 67-81% of customers prefer self-service options (Forrester, Zendesk), only 9-25% successfully resolve issues via self-service alone (Gartner). For event contexts with immediate, location-specific needs, we estimate 20-30% of queries would have required staff assistance.
Based on consumer preference studies showing majority prefer to search independently before asking for help. However, Event Copilot provides value through curated, real-time answers vs. fragmented search results.
Research shows significant friction in help-seeking behavior. Many attendees would abandon the search or resort to trial-and-error navigation rather than wait in line or search extensively.
Percentage of conversations resolved with 1-2 messages, indicating the AI provided satisfactory answers without extended back-and-forth. The 11.3% with 5+ message depth likely involved complex, contextual follow-ups that websites cannot handle.