Hotwire.com Revenue Management

Darren Koch presented on Hotwire.com’s use of ILOG business rules in revenue management. Summary:

• Ongoing segmentation and optimization help businesses serve customers
• Smart testing + flexibility = better service = higher profits
• Continues to show ROI that is increasing over time

Hotwire.com was founded in 1999 to help travel partners (who invested) sell excess inventory without driving down prices over all. Now part of Expedia (with TripAdvisor and Expedia.com), 9th largest travel site and focused on short, last-minute trips. Revenue management is a challenge for Hotwire.com as suppliers don’t want to have cannibalization and want to maximize their revenue while offering a discount price. From a web perspective they have to sort results to expose best deals while managing the fact that deals are “opaque” in that consumers don’t know which hotel is in fact represented – must generate trust but also use price markups that are optimal.

Originally their revenue management process took intuition as to what might work, did some reporting, developed a challenger for price markup approach in Excel and stored the pricing model in an Oracle table. This table grew exponentially and became a real problem as it was a fixed structure that took months to make changes. Also could not improve sorting of orders. Using business rules allowed rapid change to customer behavior as well enabling real-time optimization. Now smaller and smaller segments are developed to incrementally improve the results. They use adaptive control – champion/challenger testing – to test new approaches on small samples before putting the more fine grained strategy into production for everyone.

Quickly found pros and cons. Pros:

• High ROI – met goals for total ROI half way through 6 month project
• improvements and returns are continuing to increase
• Fast response – so much so that this group gets asked first when a change is needed
• Flexible, efficient, cost-effective

Cons:

• Complex object model and rules can interact in complex ways
• Some specialized skills required
• Organizational concerns
• Technical and business risk
• Some upfront investment

Revenue management team now uses SAS and does complex predictive modeling. When these models offer improvement it is easy to figure out which rules are required as a result and ILOG allows them to implement the rules quickly and accurately. Improvements seen:

• Pricing improves through more flexibilty in defining segments, time to market is faster
• Can now do rules-based sorting based on statistical models and this makes a real difference to customer behavior
• Rapid reaction to product and business changes – real business agility
• Also adding features like scoring inventory for later merchandising
  Inventory changes so fast that can only know it when someone asks. Marketing must “guess” what kinds of things will be available and now use the rules to optimize this and to display the results on the site.

The project took 2 business/3 IT people for 6 months with a couple of weeks technical consulting. 2 business analysts now do everything to do with the rues and 1 IT person to handle site/rule engine integration and synchronization.