louisrosenfeld.com logotype

Home > Bloug Archive

Jun 09, 2010: Globalizing an information architecture

I'm working with a large client who is about to jump with both feet into that twilight zone of information architecture: designing for an audience that is geographically, culturally, and linguistically global. It's a huge challenge, as anyone who's worked on such a project will attest.

I grappled with this issue six years ago while consulting for another multinational client, blogging about it here, here, and here (these posts are especially notable for the incredibly wise comments). Unfortunately, I've not had much opportunity to work in this space since then. In the interim, I was hoping that some brilliant cosmopolitan information architect (Peter van Dijck? Livia Labate? Jorge Arango?) would have figured this space out, but those folks are wicked busy. After some limited searching, I'm not sure we're that much further along than we were in 2004.

So in the interest of resurrecting a six-year old conversation, here are some questions. I plan to use these to get my client to think strategically about the challenge of developing a multi-lingual, multi-cultural, and multi-regional information architecture. Being the information architect that I am, naturally I categorized them. Please chime in with your own suggestions...


  • When a user visits an organization's web site, should he expect to access content in his native language?
  • Which languages are most common to users?
  • Does that organization operate in its own "master" language?
  • Which languages have the greatest strategic value to the organization? Which would simply be nice to support? Which aren't a priority at all?
  • Does the information architecture's native language (e.g., label lengths might be quite different, or might not translate at all) translate well into other important languages? Are there other semantic issues to consider?


  • Are there geographic regions (e.g., countries, states/provinces, municipalities) that are important to users? Do they self-identify by any sort of geographic region when they use the organization's site, or does it matter at all to them?
  • Are there geographic regions (e.g., countries, states/provinces, municipalities, sales territories) that are important to the organization?
  • Why are these important (e.g., legal issues, tax collections, sales territories)?
  • What geopolitical disputes (e.g., China and Taiwan) might impact the information architecture?


  • Are there cultural issues that impact how users interact with online content in general?
  • Are there cultural issues that impact how users prefer to interact with this specific type of organization or industry in particular?
  • Does the existing information architecture's structure work well in other cultures (e.g., one culture might value hierarchical breadth over hierarchical depth)?


  • Are there intersections of any of geography, language, and culture that stand out and merit special attention (e.g., Quebecois, Malay-speaking Singaporeans)?
  • How does the organization define the objects where these issues intersect? As "locales"?
  • Conversely, how does it define "locale" (the standard term for such an object; typically a pairing of language and country)?

The organization

  • How is comprehension of the organization impacted by language/geography/culture?
  • How about its products? Its services?
  • What resources is it currently devoting to i18n and l10n? And where have these efforts begun (e.g., with content management)?
  • Is there an emphasis on i18n rather than l10n, or vice versa?
  • Does the organization consider a locale the same as a market? How are the latter defined today the organization?

The organization's content and information architecture

  • Are there specific products, services, and/or content areas that merit l10n? Put differently, are there certain products and services that are especially relevant for specific locales? Or not?
  • How might these be prioritized?
  • How do users find their way to an appropriate locale (or how should they)?
  • Where should they find their way to an appropriate locale (e.g., main page, landing pages, critical pages deep in the site)?
  • Are there other areas of the site that should make clear that different language/culture/geographic options are available, and provide those navigational "switching" options?

email this entry

Comment: Dave Mulder (Jun 9, 2010)

Obviously, a huge challenge.

In my own study of the cross-cultural literature, I haven't found much evidence to suggest significant differences in the way that people interact with a design.

There will be, however, differences in the types of devices people use to access the web. I don't necessarily see that listed as a consideration here.

And I've seen & heard some work that suggests a preference toward localized versions (including some localized content) where practical.

These kinds of projects are way interesting. I hope you're able to blog about it moving forward!

Comment: Walter Underwood (Jun 10, 2010)

How about "what time is it?"

When something is published in Australia and someone in France searches for things from yesterday, what happens?

Comment: Hanan Cohen (Jun 10, 2010)

Quick thoughts

Politics of border representation in maps of areas that are in conflict.

Multiple languages per country US/Spanish, Kenya/English

Comment: Jesús Bustamante (Jun 10, 2010)

Having worked for a number of years in the context of the EU, I would like to suggest the "political" category to your list of questions. Politics of language are a big issue in this context. See for example the list of languages sorted alphabetically by ISO code in the default page of http://europa.eu/. I think that
at least the question if there are any specific political considerations should be asked.

Comment: epc (Jun 10, 2010)

My experience is 10+ years out of date and is more techie focused, so with those caveats:

• We tried (and mostly succeeded) to impose a top–down I/A on an organization’s “world wide” web presence but ran into frequent problems because much as the organization thought it was a top–down command–and–control we–dictate–you–follow sort of place, it was more like a federation of entities generally moving in the same direction with many regional, ethnic, nationalistic and linguistic quirks. Many thought creating a unified web presence/identity was the same as creating a unified brand identity.

• How does your systems architecture impact your I/A and vice versa? We thought they should be (and were) independent, but as different regions/countries impose different data collection and retention requirements you may find you need to physically separate systems targeted at one country or trade bloc from another.

• Should one locale be able to view the content from another locale? If the company is truly global and all products are available with minimal price differences in all markets, no problem, right? But we ran into the “problem” of customers in one locale finding out about products available in another locale which either weren’t available locally, or were but at much higher prices. This isn’t an I/A problem, per se, but an organizational process problem. But the expectation is typically that one can solve this with a good I/A and a good webmaster.

• I never personally found a satisfying solution to how to present the home page: at one point we detected the language of the browser (either checking the Accept-Language header or by guessing based on the IP) and customized the navigation and date/time references by language and the guessed country. But this just served to highlight that we had very little “global” content in anything but English. And while we frequently got beat up over this by other parts of the company, none would fund/help out with translating content. The organization in question now checks a cookie and redirects to a local country version of the homepage based on the content of the cookie, defaulting to the US if no cookie is present.

• Someone’s got to pay for it all. May not seem to be an I/A issue, but time and time again we ran into the problem that the picture the organization wanted to present online was based on a fantasy view of how the company actually worked. And the same people who pushed forward the fantasy couldn’t quite grasp that even if the “Chairman” demands it so, doesn’t make the budget lines appear or the cross-border funding transfers occur. This seems like the third variation on the theme but: is the organization prepared to support a global I/A with its internal business processes? And if not today, will it put in place a plan to update the processes? Because otherwise a global I/A is just a fantasy.

• Multiple languages was mentioned already, but one thing I’d add is that the language the site visitors use may not be the one the organization traditionally markets or communicates in. In my case, we stunned the marketing powers that be with the information that a good percentage of our US audience had specified Spanish as their preferred language. While we had marketing material in Spanish, it wasn’t intended for the US market.

• A variation on the locale vs locale problem: sometimes customers from Locale “A” end up using the site from Locale “B”. While it’s ok to default to Locale “B”s options, do you constrain the user to *only* accessing the Locale “B” site or do you allow her to break out to the “A” site.

• I initially decided to consolidate our 1000+ sites into regional sites using www.CC.example.com where CC was the country code or a locality code designating where the systems were housed. Bad, bad, bad move. People within the organization were amazingly sensitive to the use of country codes in the domain name. Even though the same information was being served off the same system in the same country under the old URL (www.ourdivisionhasawebsite.example.com), moving it to www.uk.example.com/ourdivision/ ignited a firestorm. And before I get ridiculed for American naïveté, this was something that was signed off by my international partners.

• Even worse was the replacement, instead of segregating sites under a CC.example.com domain system, all of the sites got moved “up” to just www-NNN.example.com (where NNN is a number). Problem with this? Bleeding of cookies and authentication information between the company’s intranet and internet.

I don’t envy your task. I found that as an American it was just assumed that I had no idea of what I was talking about and as time went on I found it easier to have my international partners propose the solutions we’d agreed upon in private. It eliminated the knee–jerk reactions and let us focus on the real problem (trying to project this unified global identity from a federation of 100+ separate companies).

Comment: arne (Jun 10, 2010)

Very relevant questions and very good comments, too! Thanks for sharing this topic!

I just have a few additions:
- how do you deal with RTL-locales with regards to the templates used? (the best approach seems to be a mirroring of the template, but this of course can result in significant extra work which may or may not correspond well with the width of the audience you wish to reach in these locales)

- regarding your question of emphasis on I18N or L10N I agree that from an organisational standpoint it's probably possible to find a general tendency that reflects the organisational culture (-> dominant central leadership vs. a lot of local freedom).
In the resulting IA I am convinced that this question should not be answered for a site as a whole. It will be important to identify segments and functions for which a high level of global standardization will work well from a user perspective (-> and will be more efficient to create and maintain globally), but also identify segments and functions for which a globally standardized approach will lead to insufficient local UX. In these cases our approach will typically be to consider as many modularities and local variations up-front (-> such as catering for varying address conventions and making them "switchable" in the backend), but also allow room for dedicated local developments where needed (e.g. activities relating to local promotions).

- another important aspect will be to look into the processes of localization: will localization be handled by local offices/teams? which tools/permissions/rights do they have? Which of there activities need central/global approval? How do you handle media-assets, outgoing-links and other locale-specific assets that the local teams might want to use?

- SEO can be difficult to handle globally, too.

Comment: Daniel Szuc (Jun 10, 2010)

Hi Lou, one thought is the "language of selling" and how products are presented differently in countries, regions etc

This has been one of the main sticking points in research we have conducted for multi nationals. People may just have a different way of browsing product and buying online versus an approach that may work well in the USA (as an example)


Comment: James Kalbach (Jun 10, 2010)


you're asking the right questions, for sure. Here are some more thoughts, using your categories:

- You'll find some semantic differences, so you might want to allow for localization of the IA itself. At LexisNexis we made our topic level categories of a taxonomy universal, but lower levels are adaptable.
- You also have to worry about character sets, depending on the languages you're handling
- Stop words for search won't be universal. "The" in French means "tea".
- Search connectors, like AND and OR, may need to be translated or not.
- Search relevancy is also going to be different. Make sure you're search engine can handle different languages and can integrate with the local IA properly (e.g., with a word wheel or synomyms)
- Label length is a concern, but so is wrapping. A French term may have 3 small words that make a single word in German
- Gender and endings of words have to be considered too. This is more for UI labels.
- For UI design, don't put links or controls in the middle of sentences.
- Capitalization rules are different in different languages

- There may be laws and regulations you have to consider, like the Canadian gov's requirement for 2 languages. There's also different accessibility laws.

- We've not found huge differences in culture on things like structure and hierarchy. The differences are more on a detailed level around labels and semantics.
- Some concepts don't exist in some cultures, or have special meanings.

The Organization
- I've actually found that on most of our international projects, the biggest differences are with the local business units and stakeholders. Everybody says "we are different", and to some degree the want to be different. As a results, minor differences between regions may get exaggerated and the real differences ignored.

Differences in behavior and experience seem to be more salient than language and geography, in my opinion. A power searcher in Hong Kong probably has more in common with power searchers in London and Chicago than a novice searcher in the next cubicle, for instance. So I'd recommend a clear segmentation across the board to begin with the focus your IA efforts.

There's a lot on internationalization in my book, Designing Web Navigation. See the sidebars throughout. http://www.amazon.com/Designing-Web-Navigation-Optimizing-Experience/dp/0596528108

I also touch on some of the challenges we've had a LexisNexis in a presentation given at the 2006 IA Summit: http://www.asis.org/~iasummit/2006/conferencedescrip.htm#107

Comment: Lou (Jun 10, 2010)


Fantastically useful comments, and, not surprisingly, fantastically generous commenters... Thanks so much! Still processing what you've all said (and glad for more suggestions, of course!).

Comment: Glamintern (Jun 11, 2010)

Needless to say, getting out there and doing primary research puts things into perspective. I've found that even if one can't physically travel to all regions for whatever reason, phone interviews can really help. Don't stand down on the dedication to research because it is more difficult/expensive etc. Think laterally and find a way.

Comment: Eleonore (Jun 11, 2010)

Hi there,

While I'm a communications professional and fairly new to the information architecture field, I've created the architecture for a resource website for Aboriginal community workers, to support the work that my organization does in Canada.

I consulted with an Aboriginal advisory committee, and learned interesting things about culturally appropriate interface that I thought might be appropriate to share here.

I think that, when a website is targeted at a community with a strong cultural identity, particularly a minority where the inclusion of cultural elements in communications is regarded as a vital part of the preservation of the culture itself, it may be important to "brand" the website in a specific way. I agree with others here, however, that the structure itself does not seem to be impacted by the culture of the user, although I'd be curious if there are some examples of this.

When working with First Nations communities, I was told that using the colour red and circular and rounded shapes would make the user more comfortable when browsing. However, Inuit and Métis users explained to me that their cultures did not identify as strongly with the colour red as with the colour blue, so I integrated both elements. Furthermore, languages (in this case, Inuktitut and Cree) were desired, even if most users would speak English or French as well. I think that there's historical context to this - integration of cultural/language elements on online spaces is also a validation of the various Aboriginal cultures.

Any thoughts?


Comment: Richard Dalton (Jun 12, 2010)

I've always approached design problems that include satisfying multiple audiences (which is essentially what a globalization problem is) by first doing research or design iterations for each distinct audience type (locales in your case) and then seeing whats similar and can be shared vs. what needs to be different. This degree of difference will determine whether a single site can be the foundation of the experience for all the audience types or if you'll need several, significantly different foundations and sites.

Comment: Fil Sapienza (Jun 14, 2010)

You may be interested in research I conducted about multi-ethnic card sorting (see link above). I also have some empirical studies from Russian-American and Latino / Mexican-American users on the site.

Comment: Cliff Tyllick (Jun 28, 2010)

Wow! I'm bookmarking this for more careful reading when I can devote more time to it.

But you're right to question all assumptions about language. For example, at the UPA conference in Munich last month, Markus Weber made the point that German usability practitioners use English terms even when communicating in German. If you were to deliver a presentation in German and translated "usability," "user experience," and similar terms of art into German, your audience would have a hard time following you, because they would have to mentally translate your German terms into English to be sure they knew what you meant.

So do question the assumption that everything needs to be translated. Perhaps this customer's field has adopted the terms in one language or another as its terms of art, and trying to translate them would only create confusion.

Add a Comment:



URL (optional, but must include http://)

Required: Name, email, and comment.
Want to mention a linked URL? Include http:// before the address.
Want to include bold or italics? Sorry; just use *asterisks* instead.

DAYENU ); } else { // so comments are closed on this entry... print(<<< I_SAID_DAYENU
Comments are now closed for this entry.

Comment spam has forced me to close comment functionality for older entries. However, if you have something vital to add concerning this entry (or its associated comments), please email your sage insights to me (lou [at] louisrosenfeld dot com). I'll make sure your comments are added to the conversation. Sorry for the inconvenience.