Revenue Management in a Volatile Marketplace. Tom Bacon Revenue Optimization. Lessons from the field. (with a thank you to Himanshu Jain, ICFI)

Size: px

Start display at page:

Download "Revenue Management in a Volatile Marketplace. Tom Bacon Revenue Optimization. Lessons from the field. (with a thank you to Himanshu Jain, ICFI)"

Damon Griffin
5 years ago
Views:

1 Revenue Management in a Volatile Marketplace Lessons from the field Tom Bacon Revenue Optimization (with a thank you to Himanshu Jain, ICFI) Eyefortravel TDS Conference Singapore, May

2 Outline Objectives Real world environment Case study Alerts based RM approach Implementation and results Conclusions Q&A 1

3 Objectives of this presentation Reflect on reality in the field Make RM tools more effective for the end users Demonstrate implementation of simple and adaptive RM solutions Validate substantial impact of such solutions 2

4 What does the real world look like? Spectrum of RM efficiency at airlines Multiple power users of RMS Well-functioning RM system Dedicated OR staff RM driven organization Lack of analytical talent RM system out of calibration, seen as a black box Importance of RM is ignored, strategy driven by sales and marketing 3

5 RM System Maintenance Some companies lack talent to properly execute RM Ignore System; and importance of RM overall Complex system out of calibration; loss of confidence 4

6 RM System Maintenance Some companies lack talent to properly execute RM Ignore System; and importance of RM overall Complex system out of calibration; loss of confidence 5

7 Airline Forecast Process A 50 aircraft fleet can result in 10 million forecasts each night! * Marketplace * Your Commercial Initiatives * Competitor s Commercial Initiatives RM System Updated Nightly w/ Bookings Forecasts for - Each O&D - Each Flight Days out price points 6

8 Volatile environments throw off otherwise well-functioning RM systems in the absence of proper maintenance Network modifications or shocks Dynamic Competition Shifting strategies Uncertain markets: events, changing characteristics System may not react quickly enough Garbage in Garbage out Mergers, changes to fare map, etc, etc 7

Volatile environments throw off otherwise well-functioning RM systems in the absence of proper maintenance Network modifications or shocks Dynamic Competition Shifting strategies Uncertain markets:

9 Volatile environments throw off otherwise well-functioning RM systems in the absence of proper maintenance Network modifications or shocks Dynamic Competition Shifting strategies Uncertain markets: events, changing characteristics System may not react quickly enough Garbage in Garbage out The challenge becomes to implement RM techniques that can be sustained by the airline Mergers, changes to fare map 8

10 CASE STUDY 9

Case study: An airline with limited capabilities Sigma Airlines Mid size carrier ; 800 flights a week; mixed aircraft fleet RM system severely out of calibration: Significant changes in the network

11 Case study: An airline with limited capabilities Sigma Airlines Mid size carrier ; 800 flights a week; mixed aircraft fleet RM system severely out of calibration: Significant changes in the network Major issues in forecasting System too complicated for the users to fix Flight controllers were doing manual overrides on more than 2/3rd of the flights. Flight controllers claimed to know everything about their markets- we have managed flights for the last 30 years 1 Original carrier identity not disclosed 10

12 When the 70/30 principle turns to 30/70 (or worse) A large number of flights were being controlled by manual overrides with no analytic decision tool Limited capabilities to fix and maintain the complex system Required to act quickly to contain the losses: Every day the airline was missing the booking window for future flights Highly bureaucratic organization- an ideal solution could take 6 to 12 months to be approved and implemented Temporary solutions or training wheels were needed to bridge the period of volatility 11

13 APPROACH AND IMPLEMENTATION 12

14 A simple, analytical approach grounded in data 1 RM Diagnostic Analyze past and future flight departures (from RMS) Forecasts, bookings, recommendations, overrides Identify the most critical issues 2 Build alerts by time intervals Cluster analysis based on market characteristics Thresholds derived from average booking curves Controllers to still make their own decisions- special events, etc. 3 Partial automation Turn some alerts into automatic recommendations Feedback from controllers 13

15 Categorization of flights network-wide BOOKING PATTERN Early Late LOAD FACTOR High Med Low 14

on Cluster Dendogram Booking curves Load factor A-ha moments and this

16 Cluster analysis: categorize flights based on a range of market characteristics Clustering analysis used to categorize flights based on Cluster Dendogram Booking curves Load factor A-ha moments and this makes no sense Further qualitative simplification and reduction of categories 15

17 Categorization of flights network-wide: Economy Cabin Unique Characteristics % Category 1 Category 2 Category 3 High load factors, higher upsell opportunity Early booking, medium to high load factors, moderate upsell opportunity Late booking, top fares not selling 10% 44% 46% 16

18 Categorization of flights network-wide: Business Cabin Unique Characteristics % Category 1 High Load Factor in Business Cabin 27% Category 2 Low Load Factor in Business Cabin 68% 17

19 Sample rules or thresholds for closing classes to achieve upsell 100% 90% 80% 70% 60% 50% 40% 30% 20% E-3 E-4 E-6 E-8 10% 0% Alert based on booked seat factor and days before flight 18

20 Sample rules or thresholds for opening classes to build load factors 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% E-3 E-4 E-6 E-13 0% Alert to open the classes if load factor is below a threshold 19

21 Phase I: Alert reports circulated every morning Daily reports on flights with recommendations Compliance by controllers required Reports tended to drive controllers to open flights and fill empty seats Full fare ( Y ) bookings observed for the first time on several routes 20

22 Phase II: Alerts incorporated in the system Alerts appear as flagged flights in controller queues Previously, workflow revolved around forecast-based queues 3 Partial automation Phase III: Partial Automation Automate some of the alerts to become default authorizations Can be further changed by controllers Free controllers time to become more productive 21

23 Refinements from observations and controller feedback Feedback from controllers helped in Adjusting the thresholds Changing the market categories Changing the time intervals Identifying exceptions Post departure analysis helped in identifying gaps Simple structure is the key; a large number of one-off unique markets, or a complicated fare hierarchy would defeat the purpose 22

24 RESULTS AND CONCLUSIONS 23

06 m YoY RASK (Unit Revenue) Improvement = USD 2.

25 Revenue management and pricing interventions boosted Q1 revenue by over USD 3.5 million 20% 15% Budget Actual = USD 1.06 m YoY RASK (Unit Revenue) Improvement = USD 2.38 m Actual: 18% RASK Improvement 10% 5% Budget: 14% RASK Improvement 0% Feb Mar 24

26 Conclusions An alerts-based RM technique can add value to a highly sophisticated RM Forecast system Particularly valuable in volatile marketplace & more basic RM team Based on an analytical approach; implemented in a disciplined manner Improves productivity of controllers Managers can use as a monitoring tool Monitor the alerts daily, performance of their controllers and discuss any issues with them Buy-in from managers and controllers No black box: they get what they see 25

27 Thank you. Any Questions? Tom Bacon 26

90% 80% 70% 60% Loss of load factor 50% 40% 30% 20% 10% 0% 0 50

28 Flown Load Factor % Broken system and random manual overrides have resulted in chaos 100% All Flight Departures for a Month 90% 80% 70% 60% Loss of load factor 50% 40% 30% 20% 10% 0% Days to departure when E-8 was first closed 29

Loss of yield 70% 60% Loss of load factor 50% 40% 30% 20% 10% 0% 0

29 Flown Load Factor % Broken system and random manual overrides have resulted in chaos 100% All Flight Departures for a Month 90% 80% Loss of yield 70% 60% Loss of load factor 50% 40% 30% 20% 10% 0% Days to departure when E-8 was first closed 30

30 1 RM Diagnostic Detailed analysis of data extracted from RMS Data was easily accessible and was moderately clean This was critical to our approach and sigma airlines proved well-equipped in this Analysis performed at the flight/cabin/rbd level for past and future departures Analyzed and compared forecasts, system recommendations, overrides and bookings Identified the most critical issues and shared with management 31

31 Booked Load Factor Booked Load Factor Booked Load Factor Simple analyses can provide surprisingly useful insights Flight# 100/Jan/DOW=1 Flight# 200/Dec/DOW=1/Business 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Economy Business 80% 70% 60% 50% 40% 30% 20% 10% 0% B-1= J B-1 B-2 B-3 Days to departure(non linear) Days to departure(non linear) Too closed Too closed analysis by route by route Route Flights Too Close % Network wide % AAA % BBB % CCC % DDD % EEE % FFF % GGG % HHH % III % JJJ % 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Flight# 300/Jan/DOW=3/Economy Days to departure(non linear) E-1= Y E-1 E-2 E-3 E-4 E-5 E-6 E-7 E-8 32

32 2 Build alerts by time intervals Selecting time intervals for building alerts Economy Cabin Business Cabin

33 Selecting booking classes for building alerts Selected classes based on following criteria Maximize the control on bookings Classes with maximum number of bookings Sufficient fare spread between classes Selected four classes in economy cabin and two in business cabin Eco. Map E-1 E-2 E-3 E-4 E-5 E-6 E-7 E-8 E-9 E-10 E-11 E-12 E-13 Bus. Map B-1 B-2 B-3 B-4 34

34 and is saving Q2 from what would have been a decline vs. last year 10% YoY Advanced Booked RASK As of 29 Apr 0% -10% -20% As of 18 Mar As of 15 Apr -30% April May June 35

35 Closing the gaps in load factors versus last year YoY Advanced Booked Load Factor points 2 April May June As of 25 Feb As of 18 Mar As of 15 Apr 36

36 Lessons learned Real life environment is volatile by nature and is susceptible to unprecedented events Users performing revenue management can make a huge impact Investment in users is as important as in systems if not more System can be only as good as the user Required user expertise increases dramatically with system sophistication It is important to match the complexity of systems with the abilities of users Expert users using a basic system: losing benefits Basic users using an advanced system: creating chaos 37

A MAGAZINE FOR AIRLINE EXECUTIVES 2011 Issue No. 1. T a k i n g y o u r a i r l i n e t o n e w h e i g h t s. America aviation

A MAGAZINE FOR AIRLINE EXECUTIVES 2011 Issue No. 1 T a k i n g y o u r a i r l i n e t o n e w h e i g h t s SkyTeam: Caring More About You A Conversation With É Leo van Wijk, Chairman, SkyTeam Pg. 10