Japanese Web Translation

Integrating Search Engine Optimization with Website Localization

Key Issues in Developing Marketable Japanese Websites

Copyright 2006 Haiyang Yang

I. Introduction

For most companies, the ultimate goal of localizing websites into Japanese is to reach the customers in Japan and create new revenues. But as the number of Japanese webpages is growing at a colossal speed, it is becoming increasingly difficult for the localized sites to obtain the web traffic they need. There are many ways to overcome this web marketing challenge. Companies can purchase banner spaces on Japanese portal sites, participate in ad programs such as Google Japan's AdSense, or even launch commercial campaigns on traditional media. One of the most powerful and cost-effective strategies, however, is integrating searching engine optimization (SEO) with the localization process. In this article, we will discuss the key issues in this integration, and explore the different aspects of Japanese SEO.

II. What is SEO and why is SEO important for Japanese websites?

Searching engine optimization is the process of fine-tuning the content, structure, coding, design, and other elements of a website so that it can be ranked the top positions in search engine results for certain query keywords. As Japan's Internet population and e-commerce market continue to grow, ever more companies are adopting SEO to gain competitive advantage. In fact, according to SEMPO Japan's survey this year, 40.7% of the surveyed companies consider search engine related marketing a top priority and another 49.5% are interested in exploring this new marketing arena.

There are many reasons that SEO has become such an important website marketing tool. First of all, search engines are the primary traffic generator for most Japanese websites. As a recent survey by Yahoo! Japan indicates, 80% of Japanese Internet users locate websites through search engines. This means that having a search-engine-friendly website is essential to reaching online customers in Japan.

Second, high rankings in major search engines directly translate into high website traffic. According to a joint research by Internet.com Japan and Info Plant Co., Ltd. this August, almost 70% of Japanese search engine users do not look beyond the first four pages of the search results. In other words, websites that have high rankings in search results receive the majority of the search-engine-generated traffic.

Third, when people are using search engines, they generally have genuine interests in the keywords they are querying with, and thus have a high potential to be converted into real customers. In fact, a marketing survey by NPD Group shows that people are five times more likely to purchase a product or service after finding the website through search engines than through banner advertisements.

Forth, not only is SEO an highly effective web marketing tool, it is often the least expensive one. As the research by Piper Jaffray & Co. indicates, the per-lead cost of search engine marketing is lower than that of any other marketing method 4.5 times lower than banner ad and 22 times lower than direct mail.

III. Why should SEO be integrated with localization?

Integrating SEO with website localization can offer financial advantages for both localization service providers and their clients. For the localization companies, the integration expands their service range and creates new revenues. For the client companies, the integration helps reduce the overall cost of launching the Japanese websites by combining localization and SEO into one coherent project. The client companies no longer need to hire a separate SEO firm after the websites are localized.

SEO and localization integration may also help increase the quality of SEO, as the web marketing elements will be thoroughly incorporated into the entire translation and localization process. Furthermore, because SEO and localization are implemented simultaneously, the integration may shorten the amount time needed for the localized sites to receive healthy web traffic and generate profits.

VI. How should the integration be done?

Before we dive into the details of the integration, it is important to understand that there are two types of search engines: human-based search directories such as Yahoo! Directory and DMOZ, and crawler-based search engines such as Google and MSN Search. For directory search engines, human editors will decide whether a website should be included in their listings by examining the quality and uniqueness of the content, level of usability, trustworthiness, and many other aspects of the site. In other words, optimization for this type of search engines is essentially building a website that satisfies the human editors.

The crawler-based engines are the focus of this article. For this type of engines, a webpage's ranking is determined by computer algorithms that are based on (1) textual information such as keywords proximity and keywords density, (2) linkage information such as the number of in-site and off-site links of a webpage, (3) traffic information such as the amount of click-throughs and visit durations. We will not discuss general SEO topics here. These topics are covered extensively in books such as David George's The ABC of SEO and Andreas Ramos and Stephanie Cota's Insider's Guide To SEO. Instead, we will concentrate on the areas related to integrating Japanese SEO with website localization: translating keywords, localizing HTML codes, modifying layout, and handling dynamically generated pages.

1. Translating Keywords

Currently, many localization teams use special software to ensure translation consistency of key terminologies, phrases, and sentences. This consistency, however, may not be ideal from the SEO point of view. When people are looking for information on a product or service, they may not use the same keywords in search engines. The localization team must take this into consideration, and make sure that the keywords targeted customers are likely to use are incorporated into the translations.  For example, if "natural soap" is the key phrase, the translators should think from the potential customers' point of view and find the popular ways of expressing this phrase in Japanese. In the context of soap product, the word "natural" is usually expressed as 天然[の] (tennen[no]), ナチュラル[な] (nachuraru[na]) or 自然[の] (shizen[no]), but some people may also use the term 無添加[の] (mutenka[no]) which literally means additive-free. The word, "soap," is usually expressed as 石鹸 (sekken) or ソープ (sopu). Thus, the localization team should consider incorporating eight possible combinations of translations – 天然石鹸, 天然ソープ, ナチュラル石鹸, ナチュラルソープ, 自然石鹸, 自然ソープ, 無添加石鹸, 無添加ソープ – into the Japanese website.

Furthermore, as Jack Halpern points out in "Issues in Japanese Information Retrieval" (Multilingual Computing & Technology #64 Volume 15 issue 4), one of the unique properties of Japanese language is its "lack of a standard, universally accepted orthography." For a single Japanese word, there can be a large number of orthographic variants that differ in kanji, hirakana, katakana, or romaji. The implication of this is that if a keyword is translated using just one variant form, users querying with other variant forms may not be able to locate the website through search engines. For example, the word, 石鹸 (sekken), has five commonly-used variant forms as shown in the table below:

	Variants of Sekken
Variant	Variant Type	Yahoo! Japan Hits	Google Japan Hits
石鹸	all kanji	8,980,000	5,910,000
石けん	kanji + hiragana hybrid	3,450,000	3,230,000
せっけん	all hiragana	1,920,000	765,000
セッケン	all katakana	128,000	55,600
石ケン	kanji + katakana hybrid	104,000	66,900

(Search engine results of September 20, 2005)

Due to linguistic and technical difficulties such as handling the massive number of homophones and performing accurate morphological analysis, many Japanese search engines do not support cross-variant querying. Therefore, localization teams should consider using the different orthographic variants of keywords in the translations so that no matter which form of variant users query, the website will always be visible in the search results.

2. Localizing HTML Codes

In order to reduce cost, some companies choose to only localize body text, navigation images, and other contents that are highly visible to the end users. The not-so-obvious contents such as the information in title tags and images' ALT attributes are left unlocalized. However, these seemingly nonessential contents may help a localized website achieve better search engine rankings and obtain more web traffic.

Because the information between <TITLE> and <\TITLE> tags is usually a one-sentence summary of its webpage's content, most search engines record this text and assign considerable weight to it in the ranking processes. This information is also the website's first line of text shown in the search results, and serves as the hyperlink to the website. This text line is arguably the best place to call targeted customers to action – click on the link and visit the website. If the text is in a language the users do not understand, however, the localized site will lose valuable traffic to rival websites that have well-written Japanese title information.

The ALT attribute of <IMG> tags is another place localization should not be overlooked. Many search engines index this alternative text and utilize it in calculating webpage rankings. Image search engines, which are another effective web traffic generator, also record this text and use it to help display relevant images to web surfers. For example, the following codes will help image search engines associate the keyword, 天然石鹸, with the image file and webpage:

<IMG SRC="../image/natural_soap.jpg" ALT="天然石鹸">

(Yahoo! Japan's Image Search)

Although many search engines no longer consider the information in <META> tags important and some even ignore the tags completely, the information should be localized if possible (particularly the information in DESCRIPTION and KEYWORDS attributes). In the fiercely competitive Internet market, a slight improvement may help a localized website receive many more visitors.

As the conclusion, localization teams should help their clients understand that although some of the information inside HTML tags may seem unimportant to the end users, it can affect a webpage's search engine rankings. It makes good business sense to have the information properly localized, as short-term savings in localization cost here may lead to long-term loss in web revenues in the future.

3. Modifying Layout

Many Japanese website localization projects run into layout problems because the Japanese translations are either much longer or shorter than the original English text. Although decreasing or increasing the font size and boldness may appear to be easy solutions, this kind of changes can affect how the website is ranked by search engines. Most search engines analyze the styles of webpage text. Text with above average size and boldness is given more weight in search engines' ranking algorithms. For example, words that are surrounded by <H1>, <B>, <EM>, or <STRONG> are considered more important than the text without style tags. Thus, localization teams must be cautious about changing text styles. Demoting the styles of keywords or promoting the styles of nonessential text can reduce the prominence of the real keywords and make non-keywords be considered important in search engines' ranking processes.

Furthermore, the worst solution to layout problems is converting Japanese text into images. Today's search engines are only capable of indexing real text; textual information inside graphic files is not retrieved. In other words, once the text is converted into images, none of the Japanese translations can be picked up by search engine crawlers.

4. Handling Dynamically Generated Pages

Because it is easy to implement design templates and backend databases with dynamic pages such as PHP, JSP and ASP, many companies choose to display their localized contents by passing a country or language code parameter to the dynamic pages. For example, a Japanese language code can be passed to a PHP page with the following URL:

http://www.aGlobalCompany.com/globalContent.php?language_code=japanese

Passing a single parameter like this is not necessarily search engine unfriendly; most search engines can index this type of URL smoothly. However, if several parameters are passed, the URL may not be processed properly:

http://www.aGlobalCompany.com/globalContent.php?session_id=1234
&referral_id=4321&product_group_id=5678&sort_method=ascending&language_code=japanese 

Some search engines may strip out the parameters in the URL before indexing the page. This is because the parameters can contain a sort order, session ID, referral ID, or something else that does not lead to new content and hampers the quality of search results. Furthermore, as dynamically generated pages can be created unlimitedly, they may trap search engine crawlers in a infinite loop.

Therefore, it is best to avoid passing larger numbers of parameters in URLs. If dynamic URLs must be used, localization teams should consider the following possibilities:

1. create extra static pages that contain important product information and are optimized for Japanese search engines
2. change web servers' stop characters, such as ?, & and =, to other characters so that dynamic URLs will appear static to search engine crawlers
3. utilize the major search engines' pay-for-inclusion programs which guarantee that the dynamic pages will be indexed properly and timely.

V. Conclusion

SEO is a powerful Internet marketing tool that can help global companies reach their Japanese customers and create new revenues. Integrating SEO with website localization not only can offer cost advantages, but also can help increase the quality of SEO and shorten the amount time needed for localized websites to receive healthy traffic. In this integration process, localization teams should spend extra energy on translating and selecting the most appropriate Japanese keywords; all popular translation alternatives and variants should be incorporated into the websites. Localization teams should not leave the information in important HTML tags untranslated because it can affect how the webpage is ranked. Localization teams should also be cautious about changing text styles or converting text into images; these modifications can have a detrimental effect on the websites' search engine rankings. Finally, because search engine crawlers may have problems with processing dynamic URLs, it is best to avoid using such URLs. If parameters must be dynamically passed, a workaround should be implemented to ensure that the contents of the website are visible to search engines.

References

1. Halpern, J. (2004). Issues in Japanese Information Retrieval. Multilingual Computing & Technology, 64, 48-50
2. Internet.com Japan. (2005). Research on Internet Search Technology. Retrieved September 20, 2005,
   from http://japan.internet.com/research/20050808/1.html.
3. iProspect.com Inc. (2004). Search Engine User Attitudes Survey. Retrieved September 20, 2005,
   from http://www.iprospect.com/premiumPDFs/iProspectSurveyComplete.pdf.
4. Ministry of Public Management, Home Affairs, Posts and Telecommunications of Japan. (2005). Information and Communications in Japan. Retrieved September 20, 2005,
   from http://www.johotsusintokei.soumu.go.jp/whitepaper/eng/WP2005/2005-index.html.
5. NTT Data Institute of Management Consulting, Inc. (2004). Survey on E-commerce Market. Retrieved September 20, 2005,
   from http://www.keieiken.co.jp/news/news11/news_11.pdf.
6. Rashtchy, S. (2004). New Methods in Search Marketing. Retrieved September 20, 2005,
   from http://www.sempo.org/research/Rashtchy-Presentation.pdf.
7. SEMPO Japan. (2005). Survey of SEM in Japan, 2005. Retrieved September 20, 2005,
   from http://www.sempo.jp/contents/19/28.
8. Thurow, S. (2005). Search Engine Optimization: What to Know Before You Build. Retrieved September 20, 2005,
   from http://www.informit.com/articles/article.asp?p=31193&seqNum=2&rl=1.
9. Yahoo! Japan. (2005). Japanese User Survey 2005. Retrieved September 20, 2005,
   from http://docs.yahoo.co.jp/info/research/wua/200504/page14.html.