Managing Localization from End-to-end - Going Global with DITA

Managing Localization from
End-to-End:
Going Global with DITA
Keith Schengili-Roberts | November 16, 2010

Localization at AMD

• Introduction
• Who We Are and What We Do
• Understanding the Importance of Localization
• Improving our Processes
• Making Localization Work for Us

2

Who We Are and What We Do

3

Who am I?

Keith Schengili-Roberts
• Manager for AMD’s Documentation
and Localization department for the
Professional Graphics division
(formerly ATI)
• Lecturer at University of Toronto’s
Professional Learning Center since
1999, teaching courses on information
architecture and content management
(sample slide decks available from:
http://www.infoarchcourse.com/)
• Author of four titles on Internet
technologies; last title was “Core CSS,
2nd Edition” (2001)

4

AMD Worldwide

Founded in 1969
HQ in Sunnyvale, California
Employs over 10,400 worldwide
Annual revenue of $5.4 Billion (2009)

5

AMD CPG and GPG

CPG = Computation Products Group, CPU focused

GPG = Graphics Products Group, GPU focused

6

End-User Documentation

My team is responsible for GPU
product documentation:
 End-user manuals for
Radeon/FirePro
products
 Online help for Catalyst
Control Center
 Custom documentation for
Dell, Fujitsu-Seimens, etc.
 Some marketing materials,
such as product packaging
and inserts

7

Engineering Documentation

We are also responsible for GPU
documentation aimed at OEM/ODM
partners:
• Technical Training manuals
• Other internal training materials
• Hardware-related specification
documents, ASIC databooks,
etc.

8

Understanding the Importance of
Localization

9

Why is Localization Important to Us?

Localization is vital to AMD for two reasons:
• It allows us to reach out to our customers
• It enables us to do business on a global scale

10

Why: Reaching Out to Customers

http://www.youtube.com/watch?v=S3r51a6_q9M

http://www.youtube.com/watch?v=S3r51a6_q9M

11

Localizing the AMD Web Site

12

Localizing AMD Marketing Materials

13

Why: Enabling Business Across the Globe

By localizing our documentation, we can:
• Provide ODMs/OEMs with detailed technical information in their
native language
• Provide end-users with language-specific GUI/documentation
• Sell products in markets that have local/regional language
requirements
• Enhance brand awareness and experience

14

Localization as a Requirement to Do Business

• In Canada, all boxed shelf
product must be in English and
French
• In Turkey, we are required to
produce end-user materials
only in Turkish
• Around the world our OEM
partners expect end-user
manuals in multiple languages
in order for them to sell their
product (typically 9 languages
including English)

Business users on the Web are three times more likely to buy from sites
in their native language and stay two times longer on sites in their own
language. What’s more, a full 76% of all Web users who encounter sites
not in their native language either look for a translated version or leave
the site.
“Are you Really Globalizing Your Website?” by Prema Nakra (2006)

15

OEM Language Requirements

16

Localization Deliverables
Currently, we localize to a maximum of 25 languages:
 Arabic  Finnish  Portuguese
(Brazilian and
 Chinese  French Iberian)
(Simplified)
 Hebrew  Russian
 Chinese
(Traditional)  Hungarian  Slovenian
 Czech  Italian  Spanish (Catalan
 Japanese and Latin-
 Danish American)
 English  Korean
 Swedish
 German  Norwegian
 Thai
 Greek  Polish
 Turkish
Taken together, this enables us to have localized products for over
80 countries.

17

GPG Software Localization Target Languages

18

Localization for HR and Legal

Localizing internal
communications allows us to
reach out to our own people in
their own language. Some
samples:
• Our “Worldwide Standards of
Business Conduct” document
is localized to 7 languages
• HR forms
• Internal training materials

19

Poor Translation Affects Branding

Some classic mistranslations:
Chevy Nova = Spanish: “Doesn’t
go”
Finger-lickin’ good = Chinese "eat
your fingers off" (not
appetizing); in Iran was
translated as "so good you'll
eat your fingers”, which is a
traditional Farsi
saying/compliment
Intel Pentium IV = original Korean
translation: “chip of death”
Examples derived from Time Magazine’s
“And What Does it Mean in Farsi?” on
localization blunders:
http://www.time.com/time/magazine/ar
ticle/0,9171,184974,00.html

20

Need for Better Terminology Management

Examples closer to home
(none done by our current
localization vendor):
“Senior Thermal Engineer” was
translated as:
“Senior Keep-Warm Engineer”
“Catalyst A.I.” translated into
Russian as:
The English is clear enough, but the “Catalyst Intelligence Quotient”
Welsh reads:
"I am not in the office at the moment. “Click OK to finish.” translated
Send any work to be translated." to:
Keeping better track of commonly used “Click OK to [become] Finnish.”
terms (“vehicles”, “residential”) using
Translation Memory might have helped.

21

Improving our Process

22

Improving Our Process

Our process for localization was an uphill struggle before our
reforms. We had inefficiencies in our tooling, internal routing
processes and in choosing localization vendors.
When rethinking our process the following factors were vital:

• Reducing cost
• Reducing time to market
• Increasing our focus on QA
• Decreasing time spent on administration
• Increasing content reuse
• Eliminating DTP and the costs associated with it

23

Initial Steps to Making Localization More Efficient

Old toolchain was costly and inefficient; we were consistently
over-budget with no improvement in sight
Decision was made to go with using DITA XML within a CMS;
chose Ixiasoft’s DITA CMS in 2006
Began production in February 2007; have published over 2,000
documents since that time
• 46% of this is in English
• 54% is in the languages we localize to (25 maximum)
Localization process streamlined, providing more time to focus
on QA than on administration or fixing formatting issues

24

DITA Advantages from a Localization Perspective

Topic-based approach is doubly efficient:
• Focus on self-contained meaning + minimalist approach (as
opposed to the narrative style favored under the old toolchain)
leads to significantly greater opportunities for content reuse
• Smaller, more easily divisible (and consistent) “chunks” of content
= better segmentation matching for localization purposes
100% reuse of original English content is multiplied by # of
languages when the exact same content is localized
• Material is localized once, and thereafter does not have to be re-
localized (barring any changes made)

25

DITA Advantages (cont.)

No DTP means that localization vendor focuses more on content
rather than formatting style
• Have observed better quality and more consistent translations from
our localization vendor as a result
• Doing DTP takes time, not just for writers but for the localization
vendor; turnaround time for localization has improved substantially
as a result
DITA XML data lends itself well to programmatic tools
• Open standard means that there is no hidden API “surprises” for
programmers
• Value-add for localization firms is in software engineering, building
software tools to automate processing on top of existing translation
workbench software

26

DITA Reduces Localization Delivery Times

In addition to greatly-reduced costs, removing DTP from
equation reduces turnaround time greatly
• In our experience, a typical 200-page doc that would have taken 3
weeks to handle using the old toolchain now takes 2 weeks or less
• Prior to using the DITA CMS, we could only hope to localize Catalyst
Control Center software every 6 months; now, we can keep up with
the monthly software release cycle.
Topic-sized “chunks” lend themselves to parallel processing
• A team of localizers can more easily be tasked to work on batches
of discrete topics than a more typically “narrative” work would
Pro: faster turnaround time
Con: consistency problems can arise; this is where additional
software tools and processes (provided by the localization
vendor) can greatly help

27

More Nimble Localization Processes

DITA XML allows us to be more nimble: for critical large
documents, we can send the localization firm finished “parts”
as we get them (“70/20/10”):
• When roughly 70% of a large document is done, we send it off for
translation, which is followed a week or two later with another 20%
of new and updated material, then the final 10% when the source
document is completed.
• While this process does cost more than sending in a whole
document at once, it reduces the turnaround time from weeks to
days, and quality is much improved because it is not done in a
rush.
• This approach was simply not feasible using our old toolchain;
ultimately, the new toolchain is still cheaper and much faster.

28

Making Localization Process More Efficient

CMS ROI
“Bad Old Days”

Content audit +
Single-sourcing

Blue line= localization budget for quarter Red line= actual localization spend

Our annual localization budget is now 2.5 times less than the year before we
started using the CMS (2006)

29

Relationship Between English and Localized Publications

30

Localization Segments Auto-translated within CMS Monthly

Portion in orange is
the percentage that
were 100% matches,
and never needed to
be sent for
localization

From July 2008 –
July 2009 an avg. of
54% of segments
were auto-translated
within the system

31

Making Localization Work for Us

32

Searching for a Company-wide Localization
Vendor
Prior to this year there was no coordination for localization purchases
within AMD.
An internal survey of localization purchases made in 2008 found that
we were dealing with over 30 individual localization firms. The
following problems became immediately apparent:
 Translation Memories (TMs) were not shared, so the quality of
translation was highly variable.
 In many cases work was obviously sub-contracted, leading to
higher-than-necessary localization costs.
 Processes across various regions tended to be haphazard.
 Regional offices were often locked into using a single regional
vendor.

33

Selecting a Vendor

A team was convened containing members from Marketing, Sales,
Engineering, Legal, and HR to advise on the firm’s localization
requirements.
The selection process had two distinct phases:
• Request for Information (RFI): The first phase winnowed out
vendors who could not meet our needs.
• Request for Purchase (RFP): The second phase further pared down
the field and ranked the remaining vendors on TQCFS principles:
Technology (processes, hardware/software used), Quality, Cost,
Flexibility, and Service.

34

Requesting Information

Gathering information is the first stage of the process. We had
to determine who could suit our needs. The following steps
illustrate how we got the search started:
1. Contacted existing vendors that we had an ongoing relationship
with
2. Questions were devised in consultation with key members of
divisions within the company to determine minimum localization
requirements
3. Consulted with Common Sense Advisory group
(www.commonsenseadvisory.com) at an early stage to help us vet
the validity of our RFI questions
4. Included most of the “Top 10” firms (for a recent Top 25 list see:
http://www.commonsenseadvisory.com/Research/All_Users/080528
_QT_2008_top_25_lsps/tabid/1492/Default.aspx)

35

Typical RFI Questions Asked

• What are the locations, if any, of your regional offices? (regional
service)
• Are localization/translation services your firm's primary core
competency? (core competencies)
• Within the last 24 months, how many Fortune 100 companies has
your firm worked with? (experience, trust)
• Does your company have experience in the
IT/Semiconductor/Graphics industries? (experience in domain)
• How would you rate your firm’s experience with specialized
localization services for: Marketing/PR Material/Legal/Technical
Publications (ability to handle all localization needs for AMD)
• For 2009, what is your year-on-year growth rate year to date?
(fiscally stable? Long term client?)
• Can you provide translation services that utilize DITA, XML and
XLIFF? (base DITA XML/XLIFF requirement)
• Localization coming from our group comprises approx. 40% of
AMD’s needs by volume; vendor must have DITA XML
capabilities

36

Some Reasons for Disqualifying Candidate
Firms
• Missing key language requirements
• Revenues were too small (small firm), or were declining (viability
called into question)
• Failing to meet threshold for number of years of expertise in
computer hardware domain
• Did not have sufficient regional representation
• Could not meet needs for marketing/PR material localization
• Listed localization as not being their core competency
• Could not handle DITA/XLIFF
• out of 29 companies that responded, 3 said that they could not
handle this requirement
• While 10% non-compliance is still significant, on the whole
this underscores the widespread ability of most major
localization firms to handle DITA-based content (this was
certainly not the case even a few years ago)
37

Request For Proposal

The RFP phase focused on the top 11-rated respondents. They were
given a much more in-depth questionnaire, with responses held under
an NDA to encourage sharing of information.
Candidate firms were also given sample documents to translate from
several different divisions (engineering, marketing, HR, sales, legal)
which were graded by native speakers from those backgrounds within
the company.
Questions were separated into four major categories, all examining
breadth and depth of expertise:
 Probed the depth of their services
 Examined their localization processes
 Teamwork and localization personnel
 Pricing structures for various localization services

38

Sample RFP Questions

• Does your firm provide terminology management services?
(depth of service; also important to co-ordinate localization
processing)
• Do you have an in-house engineering team dedicated to setting
up/automating processes for handling new file formats?
(critical to knowing how flexible they would be in handling DITA
XML content, and any other doc types we might send their way)
• How long does it typically take to localize a 250-page document
into Simplified Chinese, assuming no translation memory?
(response provides base level of how fast a large job could be
tackled; also of credibility of response)
• What is the average employment duration of translators in your
company? (localization firm with low turnover level equals more
consistent translations)
• Please describe the training process to maintain / improve the skill
levels of your translators. (want to see evidence of continuous
improvement of quality)

39

Responses Rated on TQCFS

Responses were slotted into TQCFS categories (Technology, Quality,
Cost, Flexibility and Service) and then graded on a 7-point scale by
the members of the localization team.
Some of the key factors that helped us select a final vendor included:
• Application of up-to-date technologies to support translators’
workbench
• Experience with next generation translation processes and
techniques
• Applied solution for client notification
• Team’s skills, experience and personality profile match with needs
• Scope and quality of the QA processes in place
• Response time to inquiries
• Team motivated and willing to go the extra mile
• Local representation in key countries
40

RFP Scorecard

• Scores were weighted.
Overall quality was rated
the highest, followed by
cost, services, and
flexibility.
• We started negotiations
with the top 3 firms.

41

Conclusion of Process

Keep in mind that the vendor selected best meets AMD’s
needs; our requirements are not necessarily yours

42

Future Needs

Our current plan is to move all of our engineering documentation into
our tool chain for the following reasons:
• The more efficient our process gets, the lower our localization
costs become, which means it becomes financially feasible to
localize materials that were previously not localized.
• As processes become more efficient, it becomes possible to
consider adding additional languages within the existing
localization budget.
Working to improve the quality of existing toolchain further:
• Asked Ixiasoft to work with XTM International to integrate the
latter’s XML-based Translation Management System (TMS) with
the former’s DITA CMS
• Internal rollout expected this quarter; designed to improve
localization workflow and greatly increase segment matching
done within the system, further reducing the wordcount to be
sent out for localization
43

A Good Working Relationship Goes Both Ways

Ideally, you and your localization vendor should forge a long-term
relationship:
• Actively work with your localization vendor to provide them with
information that enables them to do a better job localizing your
materials.
• Provide updates to your localization vendor when you plan on
implementing programmatic changes to DITA XML content (such
as any specializations) so there are no surprises for anyone.
• Emphasize that you value (and will pay for) nimbleness; this
encourages process innovation on both sides!
Using DITA XML has allowed us to progress this far; much of what is
discussed in this presentation would be very different if we remained
on the old tool chain. We would not be able to take full advantage of
opportunities that our localization vendor provides for us.

44

Managing Localization from End-to-end - Going Global with DITA

More Related Content

What's hot

Viewers also liked

Similar to Managing Localization from End-to-end - Going Global with DITA

Managing Localization from End-to-end - Going Global with DITA