Travel Arrangement and Reservation Services2023Machine Learning (classification)NLPB2C
Amtrak

Amtrak's 'Ask Julie' virtual assistant increases bookings 25%, generates 30% more revenue per booking, and delivers 800% ROI

Amtrak deployed Ask Julie — an intelligent virtual assistant built by Next IT (now Verint) — to serve 375,000 daily website visitors, boosting bookings 25%, generating 30% more revenue per booking versus other channels, delivering 800% ROI, and saving $1 million in annual customer service costs.

Return on investment800%
Booking increase25%
Revenue per booking vs. other channels30% more
Questions answered annually5 M+
4 min read

Background

Amtrak's 375,000 daily website visitors needed immediate, accurate answers about routes, schedules, policies, and booking — a volume that was straining phone and email support. Amtrak wanted every visitor to have access to instant self-service while reducing live agent load and improving the booking experience.

What Was Implemented

  • Ask Julie, an intelligent virtual assistant built with Next IT (now Verint), deployed on Amtrak.com beginning 2012
  • Assists with train bookings (prefilling scheduling forms, guiding the booking flow), policy questions, hotel/car reservations
  • Serves unlimited simultaneous visitors with no wait time
  • Integrated upsell prompts during booking interactions (e.g., hotel add-ons, seat upgrades)

Results

- 800% return on investment (vendor-reported) - 25% increase in bookings - 30% more revenue per booking compared to other booking channels - 32% increase in containment (inquiries resolved without escalating to human agents) - $1 million in annual customer service cost savings - 5 million+ questions answered annually - 50% year-over-year growth in Julie's usage (book frames as "50% higher user engagement")

Lessons

  • Virtual assistants in high-volume transactional contexts generate superior ROI when they complete transactions, not just answer questions
  • Upsell prompts integrated during the booking flow materially increase revenue per transaction
  • Containment rate (resolving inquiries without human escalation) is a direct proxy for cost savings
  • Transparent AI personas — customers know they are talking to a bot — can be highly effective when the task is bounded and the assistant is reliable

Ready to implement AI in your commerce operations?

McFadyen Digital helps teams move from case study to live implementation.

Talk to an expert →