The issue with Traditional RM - EyeforTravel... Smart Travel Analytics The issue with Traditional RM...

www.simon-kucher.com

Smart Travel Analytics

The issue with Traditional RM

Amsterdam, Nov 2016

Dimitris Hiotis

London office1 Plough PlaceLondon EC4A 1DE, UKTel. +44 20 7832 6700 [email protected]

The issue with Traditional RM eye for travel.pptx

RM Bingo (very old-school RM) to look for

2

B I N G O

200

B I N G O

99

Let’s start our story with a quite sophisticated RM system

…years to develop and implement

…thousand prices changed a day

% were set at user-defined parameters

1,000s of booking curves Network optimisationPrice elastic dynamic forecasts

£500 £525 £550 £575 7

Optimal price:£1,089

B I N G O

4


0%

20%

40%

60%

80%

100%

3

If something is a black box

£1,040 Max

£1,000 Min

£1,040 Max

£1,000 Min

Price can be within..


So why were 99% of prices at pre-defined parameters

Sophisticated RM may be very clever, but if people do not get it they end up controlling it too much, overriding its sophistication

..you start not trusting its outcome

..so you start putting parameters to control it

..and your parameters become the answer

4

..and such controls are needed to avoid stories like this one

Automation needs control and human intervention – else we could end up in in unrealistic price

Robotic pricing can lead to insane price points§ Two booksellers were

setting up prices based on each other

§ As one increased the price the other one followed

§ In a period of 10 days price spiralled to $23 million

The issue with Traditional RM eye for travel.pptx 5

…and here is a story of a simpler system

Airport car-park daily pricing tool

0

5

10

15

20

25

-

200

400

600

800

1,000

1,200

1,400

-60 -58 -56 -54 -52 -50 -48 -46 -44 -42 -40 -38 -36 -34 -32 -30 -28 -26 -24 -22 -20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0

PriceOccupancy

Lead timeActual SDs Target Price




0

5

10

15

20

25

-

200

400

600

800

1,000

1,200

1,400

-60 -58 -56 -54 -52 -50 -48 -46 -44 -42 -40 -38 -36 -34 -32 -30 -28 -26 -24 -22 -20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0

PriceOccupancy




Simpler models can be quicker, as effective and set you on a path for more sophisticated RM later


0

5

10

15

20

25

-

200

400

600

800

1,000

1,200

1,400

-60 -58 -56 -54 -52 -50 -48 -46 -44 -42 -40 -38 -36 -34 -32 -30 -28 -26 -24 -22 -20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0

PriceOccupancy


B I N G O

20%

B I N G O

6%

…weeks to develop

…increase in occupancy

…increase in revenue

B I N G O

8


RM capability should be built in steps with Traditional RM setting the foundations for later sophistication

Traditional RM Sophisticated RM

Simple Price differentiation

1

Rules based RM

2

Forecast based YM

3

Network optimisation

5

Customer choice based YM

4


Peak Peak Peak

-

50

100

150

200

250

300

-

2

4

6

8

10

12

14

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Hour of day# taxi rides B2C waiting time

Simple price differentiation is your first step


Waiting time Demand (indexed)

Off-peak(-25%)

Off-peak(-25%)

Off-peak(-25%)

§ Taxi provider had flat-pricing across the day

§ This led to high waiting times at peak periods of demand

§ …and low waiting times at off-peak periods, where demand was not in sync with supply

§ Solution:§ Introduce peak/off-peak

pricing§ Peak +5% than today§ Off-peak -25% than today

§ Result:§ Peak – better waiting times

maintaining demand§ Off-peak increase in demand

& revenue

Project Example

1

10

A rules based price approach naturally follows

Lead time based differentiation

10%

20%

35%

50%

60%

70%

80%

OTD (on the day)

Level 7

Level 6

Level 5

Level 4

Level 3

Level 2

Level 1% of capacity or #seats

£ Price points at each level

Close out (in days to departure)

a

b

c

£10

£17

£25

£35

£45

£54

£69

£85

7 2 1

Used by train operators, coach operators and traditionally by low-cost airlines

Three key levers that essentially revenue manage sales

§ % occupancy – have price build-up based on capacity utilisation

§ Price points – Start low and go-up as you move closer to departure

§ Close out rules ensure low prices are not available close to departure

a

b

c

2


Project Example

11

Below is an illustration of how we set-up a rate of sale forecast for a tour operator

A forecast-based system requires booking curves to generate forecasts based on market dynamics

Days to departure

Main components:a. Booking curve that shows expectation of demand at current price

b. Sales to date

c. Target (capacity)

Rate of sale forecast (last 4 weeks example)

§ Expected 10 pax booked in the last 4 weeks

§ Instead I got 16 pax booked

§ Rate of sale is 1.6 (16/10) times what I expected

§ From now till departure I expect to get 50

§ If I apply this recent of rate of sale expectation is 50*1.6 = 80

§ This will translate to a forecast of 56 booked + 80 expected = 136

NOTE: Typically 2-3 rate of sale forecasts are created, each with different base (e.g. last day, last 7 days, last month) and a weighted avg is used to generate a forecast

a

b

c

0180

100

50

TodayToday –4 Wks. Ago

40

56

1016

130

booking curve

actual

forecast

Expected in the last 4 weeks

Booked in last 4 weeks

a

b

c

Project example

3


Forecast based system can be combined with margin rules to drive pricing decisions

Resort Egypt – Sharm

Budget month 2014_08

Budget week Week 34

Flight MXP-SSH

Day of week Saturday

Current occupancy 91%

F’cast occupancy 197%

Seats capacity 100Hotel Capacity Fore-

castContrib-

utionMargin

Pax booked

Total pax forecast

Forecastpax (to come)

Apportion Comula-tive build

up

% build

up

Pricedecision

Hotel 1 5 20% 967€ - - - 0% 69 91% DOWNHotel 2 10 0% 746€ - - - 0% 69 91% DOWNHotel 3 272 98% 597€ 45 76 31 39% 100 132% UPHotel 4 5 64% 534€ 2 2 - 0% 100 132% STAYHotel 5 30 68% 520€ 2 6 4 5% 104 137% STAYHotel 6 30 120% 340€ 20 65 45 56% 150 197% UPHotel 7 50 80% 120€ - - - 0% 150 197% UP

69 150 81

A

B C

§ System produces forecasts for each flight on the programme as a % of capacity (>100% flight will fill out) based on current rate of sale.

§ Equally, system produces forecasts for the hotels that will be sold with each flight, and the resulting margin per passenger each booking could bring

§ Based on forecasts and expected margin per person, rules based engine recommends a price change (e.g. UP – put price up; DOWN – put price down; STAY: Leave price as it is)

A

B

C

Forecast occupancy of flight à>100% flight will sell out

Forecast occupancy of hotel and expected margin per passenger

Rules based pricing based on forecasts of hotel and flight

Project ExampleTour operator

3

System produces a forecast for both the flight and the hotel components of package

Based on forecast of flight & hotel, price recommendation is made

Crucially, margin is also used to determine where to best increase price when relevant stock is constrained (e.g. flight capacity)


A customer choice model enhances forecasts with insights from external market


Rate architecture Competitor pricesForecasts of demand (based

on RoS & booking curves)

Sales to date (pax, margin, price)

Demand data(searches)

Availability(failed searches)

Traditional RM elements Enhancements

4

Forecasts are enriched by a selection of KPIs which feed into pricing decisions:§ Price positioning vs.

competitors§ Web demand – are my

forecasts low due to low demand or high price?

§ Failed searches (i.e. do I have less bookings because my availability is restricted rather than my price is too high)


A network optimiser optimises the revenue mix based on forecast & predicted impact of pricing action

Pkg #3 Pkg #5 Pkg #6Pkg #2 Pkg #4Pkg #1

Price

15 50 80 100 120

450400

350300

250200

CAPACITY

Pkg #3 Pkg #5Pkg #6Pkg #2 Pkg #4Pkg #1

Price

15 50 80 100 120

450400

380

360 320

275

CAPACITY

A network optimiser will compare forecast to capacity and accounting for all interdependencies (e.g. flight to hotel and hotel to flight or even between hotel nights) will recommend price changes

Optimisation is done in multiple dimensions (outbound flight, inbound flight and each stay-night at the hotel)

Margin is crucial as optimiser will tend to price-up the lower-margin products to maximise overall margin


5


Whatever you choose, you need a process, a set of reports and a team to yield manage

Mon Tue Wed Thu Fri W/E Mon

Market monitoring

Price management

Capacity changes

Demand shocks

Process Tool Decision

1

2

3

4

Daily pricing

B2S capacity changes

B2S re-balancing

Supply based pricing

Forecast based pricing

Buy

Sell

Trade

Weekly capacity management

Late deal alert

Price

Daily pricing Daily pricing Daily pricing Daily pricing

Daily pricing

Weekly capacity management

Supply change alerts

Drill Down tool



Conclusion and key take-aways

§ Sophisticated RM is clever and can be quite revenue optimal

§ However, it needs to be understandable and believable to users, as else their sophistication is wasted

§ To do this you need to develop RM capability in steps

§ First start with the traditional RM principles, from simple price differentiation to rules-based RM

§ …before you move to forecast based, Customer choice and/or network optimiser

§ And don’t forget à Process, Process, Process

§ This allows you to:- Build the right RM culture in your organisation- Find the right algorithms and methods for your business

through trial & error- Only move to hyper-sophistication when you feel there is

extra benefit


Thank you!

Rosalind HunterDirector

1 Plough Place EC4A 1DE London

United Kingdom

Tel. +44 20 7832 67 00Fax +44 20 7832 68 00

[email protected]

Dimitris HiotisPartner

Tel. +44 20 7832 67 00 Fax +44 20 7832 68 00

[email protected]

1 Plough Place EC4A 1DE London

United Kingdom

18

The issue with Traditional RM - EyeforTravel... Smart Travel Analytics The issue with Traditional RM...

Documents

Transcript of The issue with Traditional RM - EyeforTravel... Smart Travel Analytics The issue with Traditional RM...