The Web in 1998: Some Predictions

by Jakob Nielsen on January 1, 1998

International Web

1998 will be the year when the Web becomes truly international. Many countries now have sufficiently many users to sustain their own Web services, and overseas sites and users will grow at much faster rates than American ones. English will remain the leading language on the Web and Americans will still form the largest single concentration of Web users, but they will not be as dominant as in the past.

Two consequences of the increased number of overseas users:

  • All websites will need to pay closer attention to international usability: is the content understandable for non-native speakers of the language; does it need to be translated for large overseas customer groups; are icons and graphics acceptable for foreigners; do international users have difficulties interacting with the site?
  • Many sites will be designed for targeted use in countries outside North America. It is certain that a growing number of local services will be established in every country, using the local language. In addition to purely local services, there will probably also be many regional or global services, and it will be interesting to see who will get to dominate this sector:
    1. Overseas websites might expand beyond their original country to cover large regions or go global. Their advantages would be initial knowledge of local customs and a local presence (especially in the initial phases of expanding to neighboring countries) as well as the architectural benefits of knowing from the start that their service will eventually go international (in contrast, the rest of the world is usually an "oops" in American projects).
    2. American websites could go global in a single leap. Most already have many overseas visitors, and they could build on this beginning to form truly international services. Many American sites would have the benefits of larger scale and the ability to provide comprehensive and advanced services that could be customized for local conditions with relatively little work since most of the software would have been written already.

Value-Added Web Services

Websites will realize that they do not need to do everything themselves. The Web is built on linking, and the Internet is ... well ... a network. These technologies are a perfect match for letting other sites handle services that you don't want to do yourself. Two examples that are already in place are outsourcing the acceptance of credit card payments and having a discussion forum hosted on another site. Currently, most large websites install their own search engines, but it would be easier to handle search through a link to an external search engine that was maintained by search experts but could still be configured to display the search results on pages with the desired look and feel.

As another example, all corporate websites need to give visitors directions to headquarters and other company facilities. There is no need for every site to design its own maps since there are sites that specialize in mapping services. Instead, give directions through an appropriate link to a preferred mapping service. Many of these services even provide customized directions from the individual user's starting point to the desired destination. The mapping service would be paid in whatever way it otherwise got paid. Currently, this means advertising, but in the future a micropayment might ensure enhanced maps (paid by the user or by the referring site, as appropriate for the circumstances).

Unfortunately, links to many Web services currently require authors to reverse-engineer the URLs used by the destination sites. Very few sites make it easy for third parties to link to them in programmatic ways to generate desired pages. Since most websites should be interested in getting new customers referred, I encourage them to use simple linking schemes according to a protocol that is published on the site. Once specified, such linking schemes must not be changed since that would cause the referring site's services to fail, causing bad will for everybody.

In the future, increased use of XML will allow far more intelligent data interchange between sites and thus for more advanced value-added Web services.

Value-added Web services may also be marketed directly to end users. For example, home users and small businesses could get their computers backed up across the Internet. Even a modem connection will suffice for nightly incremental back-ups. Of course, the files should be encrypted on the user's computer before being let out on the Internet: the back-up service should be responsible for storing the encrypted bits safely but should not be able to read them. In general, I predict that all data will be encrypted whenever it is not being acted upon by an authorized user: certainly any time it traverses a network, but probably also when it is being stored away for later use. Such pervasive encryption will most likely not happen for a few more years because of political problems even though it would be possible to start right now since weak encryption would be sufficient in most cases.

Self-Optimizing Content and Ads

Web content will become more fluid and will adapt in real time based on usage statistics to maximize the resulting value. For example, a news site will keep running statistics of how many users click on each of the stories on the front page: those stories that prove popular will be kept on the front page longer, whereas less popular stories will be replaced quickly. It is also possible to optimize headlines: the editor can experiment with, say, five different headlines for a story and put all of them online, with each headline randomly shown to 20% of the users. After a few minutes, a high-traffic site will have collected enough data to know which of the alternative headlines attracts the most clicks, and that wording can then be shown to all users for the remaining lifetime of the story.

Advertising could also become self-optimizing. It is certainly possible to conduct real-time tests of alternative creative treatments and stay with the one that generates the most click-throughs. Because of the nature of the Web, one can track the users further and chose the ads that generate the most sales on the destination site and not simply the most clicks, though doing so might take more than an hour due to the low click-through rates for most ads and the even lower customer buying rates. The Web is simply not that suited for advertising, but at least real-time optimization can minimize losses from ads that do not appeal to buying customers.

What Will Not Happen in 1998

I predict that three much-discussed events will not happen this year: bandwidth problem solved, micropayments, Internet Explorer becomes the only browser.


Share this article: Twitter | LinkedIn | Google+ | Email