ora solomon – director, globalization

STARWOOD
TRANSLATION
ROI MODEL
ORA SOLOMON – DIRECTOR, GLOBALIZATION
©2015 Starwood Hotels & Resorts Worldwide, Inc. All Rights Reserved. For internal use only. CONFIDENTIAL & PROPRIETARY – May not be reproduced, disclosed or distributed without the express written permission of Starwood Hotels & Resorts Worldwide, Inc.
STARWOOD
WHAT IS STARWOOD?
PRESENCE
1,500 hotels globally
50% outside the US
THE SITUATION
LANGUAGE SUPPORT
WEBSITES
11 brands in 8 languages
SPG in 8 additional languages
French, German, Spanish, Japanese, Italian,
Simplified Chinese, Russian, Portuguese
Arabic, Traditional Chinese, Korean, Polish,
Turkish, Dutch, Thai, Bahasa Indonesian
Property content
Based on data driven approach and ROI
APP
3 languages on iOS and Android
Chinese (S), Spanish, Japanese
LANGUAGE TIMELINE
‘04
‘05
4 core
languages
•
Spanish
•
French
•
German
•
Japanese
‘06
‘07
‘08
‘09
‘10
‘11
‘12
3 secondary
languages
2 languages
• Portuguese
• Russian – secondary
• Italian
• Simplified Chinese
• Arabic - tertiary
• Chinese app
‘13
‘14
7 tertiary
languages
New strategy
(Phase 1) + app
• Korean
• Japanese app
• Spanish app
• Traditional Chinese
• Turkish
• Dutch
• Polish
• Thai
• Bahasa Indonesian
Core languages = all brands + all properties
Secondary languages = brand com + some props
Tertiary languages = SPG.com + few properties
‘15
‘16
New strategy
(Phase 2)
DIFFERENT COVERAGE BY LANGUAGE
SPG.com available in 16
languages
Starwoodhotels.com
and Brand.com
available in 8 languages
COVERAGE FOR PROPERTY CONTENT
Properties Translated in 2015
1,500
ALL
ALL
ALL
ALL
1,000
452
500
Core
68
34
43
31
27
22
14
9
6
2
Russian
Italian
Korean
Trad. Chinese
Arabic
Thai
Indonesian
Turkish
Polish
Dutch
Portuguese
Sim. Chinese
Japanese
Spanish
French
German
0
237
Secondary
Tertiary
DO WE HAVE THE RIGHT COVERAGE?
» How many Japanese travelers visit
hotels in Missouri?
» Should we increase coverage for
languages like Italian and Dutch?
Are we maximizing the value of our
translation spend?
CORE LANGUAGES ARE TRANSLATED
WORLDWIDE REGARDLESS OF REVENUE
e.g., Aloft Charlotte Uptown in Japanese
-- even though all 9 Charlotte properties receive less
than $20K of Japanese revenue annually
e.g., Four Points St. Louis in French
-- even though all 6 St. Louis properties combined
receive less than $5k of French revenue annually
10
SECONDARY LANGUAGES ARE SPOTTY
EVEN IN MAJOR OUTBOUND MARKETS
e.g., many Italians book NYC
properties each year
11
OUR TRANSLATION MODEL NEEDS TO
BE SCALABLE
Globalization –
Starwood Growth –
Non-English revenue share
is growing, and competitors
are investing to win…
Virtually all our brands are
tasked with tremendous
footprint growth…
THE ACTION
MOVE TO A DATA-DRIVEN APPROACH
CREATED MODEL THAT WEIGHS REVENUE VS. FULL ONGOING
COST OF TRANSLATION AT THE MARKET LEVEL (E.G., FRENCH
IN KANSAS CITY, MO)
ROOTS
(Return on Ongoing Translation Spend) =
𝑨𝒏𝒏𝒖𝒂𝒍 𝑹𝒆𝒗𝒆𝒏𝒖𝒆
𝑏𝑟𝑜𝑤𝑠𝑒𝑟 𝑙𝑎𝑛𝑔𝑢𝑎𝑔𝑒 𝑋 𝑖𝑛 𝑚𝑎𝑟𝑘𝑒𝑡 𝑌
.
/
𝑵𝑷𝑽 𝒐𝒇 𝒐𝒏𝒈𝒐𝒊𝒏𝒈
𝒕𝒓𝒂𝒏𝒔𝒍𝒂𝒕𝒊𝒐𝒏 𝒆𝒙𝒑𝒆𝒏𝒔𝒆
𝑓𝑜𝑟 𝑙𝑎𝑛𝑔𝑢𝑎𝑔𝑒 𝑋 𝑖𝑛 𝑚𝑎𝑟𝑘𝑒𝑡 𝑌
=
𝑨𝒏𝒏𝒖𝒂𝒍 𝑹𝒆𝒗𝒆𝒏𝒖𝒆
𝑏𝑟𝑜𝑤𝑠𝑒𝑟 𝑙𝑎𝑛𝑔𝑢𝑎𝑔𝑒 𝑋 𝑖𝑛 𝑚𝑎𝑟𝑘𝑒𝑡 𝑌
.
/
[𝐴𝑛𝑛𝑢𝑎𝑙 𝑈𝑝𝑑𝑎𝑡𝑒 𝐶𝑜𝑠𝑡]
− 𝑈𝑝𝑓𝑟𝑜𝑛𝑡 𝑐𝑜𝑠𝑡 + 𝑛𝑖=1
1+𝑟 𝑖
EXAMPLES OF MARKETS WITH VARIOUS
ROOTS THRESHOLDS (E.G.: JAPANESE)
<1x
3x
5x
10x
20x
50x
Japanese in
Japanese in
Japanese in
Japanese in
Japanese in
Japanese in
Charlotte, NC
Mexico
Saudi Arabia
Chicago
Los Angeles
Shanghai
9 properties
27 properties
10 properties
20 properties
17 properties
11 properties
?
?
?
The higher you set this threshold, the less markets you
translate (e.g., at a 5x threshold, you exclude Mexico)
Oahu Island, HI
Guam
New York, NY
San Francisco/San Mateo, CA
Hawaii/Kauai Islands
Maui Island, HI
Anaheim/Santa Ana, CA
San Diego, CA
Seattle, WA
Los Angeles/Long Beach, CA
San Jose/Santa Cruz, CA
Boston, MA
California North
Las Vegas, NV
Central New Jersey
Chicago, IL
Detroit, MI
Oakland, CA
Cincinnati, OH-KY-IN
Michigan South
California North Central
Orlando, FL
Alabama North
Norfolk/Virginia Beach, VA
Washington, DC-MD-VA
Philadelphia, PA-NJ
Portland, OR
Salt Lake City/Ogden, UT
Alaska
Fort Worth/Arlington, TX
Florida Keys
Houston, TX
Bergen/Passaic, NJ
Cleveland, OH
Miami/Hialeah, FL
New Orleans, LA
Greenville/Spartanburg, SC
Syracuse, NY
Hartford, CT
Denver, CO
Kansas City, MO-KS
Delaware
Riverside/San Bernardino, CA
Atlanta, GA
Nashville, TN
Iowa Area
California Central Coast
Melbourne/Titusville, FL
Lexington, KY
Dallas, TX
Connecticut Area
Indianapolis, IN
Oklahoma City, OK
Minneapolis/St Paul, MN-WI
Raleigh/Durham/Chapel Hill, NC
Memphis, TN-AR-MS
Tucson, AZ
Ohio Area
Phoenix, AZ
EXAMPLES OF MARKETS WITH VARIOUS
THRESHOLDS (E.G.: JAPANESE IN US)
Return on spend (Xx)
563
439
e.g., There are 11
Starwood properties in
San Francisco area,
with a 42x return
31
14 13
10 10 9
8
10x
8
8
7
7
7
6
6
6
6
Not shown: long tail of >60
more US markets with <2x
return (>150 properties)
46
42 40
… whereas there are 17
Starwood properties in
Dallas, but 2x return
27
23 22 22
20 19
6
6
6
6
5
5
4
4
4
5x
4
4
4
4
3
3
3
3
3
3
3
3
3
2
2
2
2
2
2
2
2
3x
2
2
BASED ON DESIRED THRESHOLD, MODEL
CALCULATES WHERE TO TRANSLATE
Overall Dashboard for Single Language (e.g., Japanese)
2014
2015
SUB-DASHBOARDS GIVE TRANSLATION
RECOMMENDATION FOR ALL MARKETS
THE SOLUTION
OVERVIEW OF DATA-DRIVEN STRATEGY FOR PROPERTY
TRANSLATION
Reinvest
savings
Selectively
reduce translation
»
Stop translating core
languages in markets where
the data shows that it doesn’t
make sense
»
By translating just ~600-800
properties each for Japanese,
French, German, and Spanish
(rather than all 1200+ today)
we still cover 97%+ of core
language revenue!
Strategically
expand translation
»
Invest in languages with
substantial upside potential –
e.g., Chinese, Italian, Dutch,
Portuguese, Russian
A PHASED EXECUTION TO LIMIT UPFRONT INVESTMENT
NOW
Former “Core” Languages
Former “Secondary / Tertiary”
All properties
translated;
no threshold
Ad-hoc
translation
Phase 1
Phase 2
Phase 3
5x
5x
5x
threshold
threshold
threshold
10x
threshold
7x
threshold
** illustrative; size of bar not scaled to exact # properties
5x
threshold
DECREASE IN CORE TRANSLATION
PUTS MINIMAL REVENUE AT RISK
AND SAVES MONEY
TODAY
Applying Threshold
# properties
translated
% revenue
captured
Threshold
# properties
translated
% revenue
captured
Japanese
1226
100%
5X
670
97%
French
1226
100%
5X
769
97%
German
1226
100%
5X
831
97%
Spanish
1226
100%
5X
847
98%
~$300k
annual savings
SAVINGS CAN BE REDEPLOYED TO
PARTLY OFFET SECONDARY COVERAGE
TODAY
Phase 1
#
properties
translated
% revenue
captured
S. Chinese
452
Russian
Phase 2
Threshold
#
properties
translated
Threshold
#
properties
translated
% revenue
captured
89%
10X
525*
7X
809
97%
68
26%
10X
91*
7X
308
72%
Italian
34
13%
10X
209*
7X
295
78%
T. Chinese
31
35%
10X
113*
7X
136
50%
Dutch
2
6%
10X
106
7X
197
62%
Portuguese
237
76%
5X
268*
5X
250
75%
~$600k
investment
*Denotes cases where properties are both added and removed; net effect shown
>$30M
Potential
Revenue Gain
TRANSLATION STRATEGY:
IMPLEMENTATION
LANGUAGES BY MARKET ARE SET
INITIAL RESULTS
Minimal loss of revenue
for core languages
No impact on
conversion rates
Initial results on
language expansion
consistent with model’s
predictions
25
QUESTIONS