I'm building a site which will allow users to enter physical street addresses. My primary target audience is US-based but I'd like the site to be global and also accommodate an international audience entering non-United States addresses. These addresses will in turn be displayed on the site.

When accepting these addresses should I allow non-extended-Latin characters in the address itself (I'm thinking Arabic/Chinese/Japanese)? Or are non-U.S. user's used to dealing with entering address info in extended latin only? Is there an obvious answer or is it nuanced?

The country is coming from a dropdown list so that's not an issue, but the city, state/province/territory, and address lines are all entered manually.

Some issues are:

  1. I generate a url slug using the city name as part of it. My current slug generation process will down convert 'ü' to 'u' (this is okay) and eliminate arabic/chinese/japanese etc. characters. I can still get a valid url but it will be a lot sparser cutting city name out.

  2. I also geocode the addresses and not sure how well this will play with non-extended-latin characters.

  3. Usefulness of non-Latin address to users who do not speak the language the address is written in.

Note: While accurate addresses are highly desirable they are not mission critical i.e., I'm not shipping/billing


To clarify, addresses are absolutely essential to the application and there is zero possibility of not including them. I just meant to say that it's really up to me what I will accept or not.

To put the question a different way, when someone in Japan orders something off Amazon.com are they entering their shipping address in Japanese characters or are they using the Latin alphabet to enter it? I'm looking for what is the standard convention for accepting non-extended Latin characters.

Are they entering:

New Miyako Hotel 17, Nishikujo-Inmachi Minami-ku, Kyoto 601-8412, Japan


新・都ホテル Japan 〒601-8412 京都府京都市南区西九条院町17

If I force them to enter in addresses in the Latin alphabet will I be going against the convention? What about if I just force the city name to be in Latin alphabet?

  • Do you have an example of the string generated? what would that string be used for? only as url? If you have a resource behind that url, then the address seems important, is important or not?
    – PatomaS
    Commented Feb 5, 2014 at 3:06
  • If you don't actually need the address, why are you collecting it in the first place? That practice is actually prohibited in the UK. Better UX would be not to require it to be entered; and it obviates the problem you ask about. Commented Feb 5, 2014 at 8:30
  • Url-encoding non-latin characters may be a better option than eliminating them. Commented Feb 5, 2014 at 10:14
  • If you don't really need them, why ask for them? It may well affect your conversion rate. Commented Feb 5, 2014 at 10:15
  • @AndrewLeach: You may "collect" addresses in the UK if you do so for the benefit of the user. Consider a map navigation site that allows you to enter locations by name instead of clicking on the map.
    – MSalters
    Commented Feb 5, 2014 at 11:56

2 Answers 2


It's better to allow non-Latin characters for two reasons: 1. For international delivery by mail, only country needs to be in Latin characters. Local post will do the job better if they'll be able to read an address in native language. 2. If address will be read by another users, it's still better to have native version, because conversion to Latin can be hard and have multiple variants (e.g. "Pushkin square" or "Pushkinskaya ploschad" for location in Moscow, Russia), thus complicating the search.

If travellers are your audience (say, booking app or travel guide) or it's likely that US users will go to worldwide locations, you can provide additional field for latin form of the address, thus serving both local and international audience.


Since you are asking from a UX perspective, your shouldn't expect users to be used to do the work, you have to do it and be prepared for the users that will input anything, with or with out intention.

If the address is not something important, then remove it and avoid all the problems related.

If you are interested on the address for some other specific purposes, launch the site with out that part, do more testing and later on, add it, plus inform the users about the change and the option to fill that information if they want.

In any case, now or later, you can restrict the way the input is coming to your application, at least from the normal users, for instance, setting every step involved to be utf-8, so if you receive something that is not in that charset, you can transform that to utf-8. Or you can notify the user back and ask for the information to be entered on a specific charset. I wouldn't do that, but it's an option.

Since the main use seems to be for the url, you can use the geocode information to do that ans skip user input, but then you have the problem of how accurate is that information. For instance, I'm like 800Km away from what Google detects due to the exit point of my provider. But you can still use that and ask the user if that's correct.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.