Doing something three years in a row is as close as we come to traditions on the Web. Having offered predictions for 1996, 1997, and 1998, I feel compelled to turn to predictions for 1999. In a fit of honesty, I also wrote a sidebar assessing how well I did in January when I predicted trends for the Web in 1998.
f course, the most important trends will take more than a year to play out. The Web has slowed down, and the time is past when dramatic changes happened every few months. The Web in December 1999 will probably look much the same as the Web in December 1998. In some way, this is my biggest prediction for the year: no big shift to multimedia or other fancy technology.
Mobile Access to the Internet
Mobile access will be the third Killer App for the Internet after email and Web browsing. During 1999, it will become common to access the Internet from portable devices with a wireless modem. Untethered use will lead to many innovative Internet services under the slogan "anyone, anywhere, anytime: connected ."
Two related predictions:
When you are online at all times and can be reached anywhere, privacy becomes precious. Users will pay extra for screening services that allow them a respite from the world. Being out of touch will be seen as a status symbol.
A portable device has to be small for users to be willing to carry it around, so Web designers have to stop designing for a fixed size screen. Instead, Web pages must scale and work on many different sizes of display. No more frequently-asked-question: "Should I design for 640 or 800 pixels?"
Resurgence of Web Standards
Netscape's forthcoming "Gecko" release is claimed to be 100% standards compliant. This is big news because Netscape always used to be the worst browser with respect to support of the official Web standards. In versions 1-4, they preferred doing things their own (usually broken) way. Much applause for this new strategy.
Even though we all love to heap criticism on Microsoft for the many things they do wrong, Internet Explorer has always deserved praise for its standards support. There is now a risk that IE will fall behind Netscape in this crucial area: let's hope that Microsoft will realize that just because they won the browser war in 1998, there are no guarantees for future success. Microsoft must drastically improve its standards support in IE 5.
I predict that competitive pressure will force Microsoft to fix the standards support from the beta to the final release of IE 5. Even if they don't, Web developers will embrace the standards and avoid any design elements that are not part of the official standards. It is simply too much hassle to maintain websites that include non-standard elements, even if said deviations cause the site to look slightly better in one of the browsers. 1999 will see the eradication of the "this site optimized for ..." blight that has contaminated the Web since 1995.
Automated Customer Service
Customer service is an area where the Web can be better than reality since customers are getting ever more dissatisfied with the service they get in the physical world: the American Customer Satisfaction Index for the retail sector has dropped from 76 in 1994 to 71 in 1998 (on a 1-100 scale).
Websites currently fail to give good customer service: not only is it close to impossible to find anything on the Web, but it is very hard to be allowed to be a customer. Smooth transactions and good follow-up are foreign concepts to most Web businesses. In particular, websites are poor at handling exceptions and odd cases, even though exceptions are the rule in human behavior. For example, I had recently bought a ticket through the website of a major airline (following my principle that you need to live a Web lifestyle if you want to be a Web pundit). When my travel plans changed, my assistant tried to rebook the ticket, but was bounced back and forth between several parts of the airline, none of which knew how to change an Internet ticket.
The simplistic answer to poor customer service on the Web is a demand for more human service operators. But this approach fails for two reasons:
support line operators are expensive and a poor fit for e-commerce businesses which are supposed to operate more efficiently than old companies
human operators drop us back in the physical world, so we miss the opportunity to give better service than was possible in the past
We can become better than reality with the insight that so-called "customer service lines" are really "complaint lines": you don't call for support unless something is wrong . A flexible Web user interface would allow the user easy access to fixing the problem without having to call anybody.
Some companies try to avoid human support operators by having a natural language system read their incoming email and send out canned responses. This is a misguided attempt to treat the symptoms rather than the disease: why are the customers sending email? Because they can't find the answers on the website! Automated email responses are no better than having people interact with a website, except that the response times balloon into several minutes per interaction. People only get annoyed with canned email responses: the expectation is that email is read by a human and not by a robot. Email does have to be minimized, but the way to do so is through a customer-centric website that contains almost all the answers in an easy-to-find manner.
Do I really predict the emergence of good self-service support on websites in 1999? It is for sure one of the greatest opportunities for ecommerce sites to distinguish themselves - both relative to other websites and relative to physical companies. Good automated service is a "small matter" of good task analysis, good usability, and some programming. Each of these three steps requires work and specialized staff, but they only need to be done once - then the computer takes over and provides the actual service to millions of individual customers. It is easier to train a computer to give good service than to train thousands of human service reps.
Even though I think most sites will continue to offer bad service in 1999, I do think that many leading sites will get serious about automated customer support. Most good sites already offer customers the ability to check their order status online: this is the tip of an iceberg that's about to hit the Web.
Web Patent Bonanza
Many patents on fundamental e-commerce business models and Web user interface improvements will finally issue in 1999. It seems to take 3-4 years for most Internet patent applications to wind their way through the Patent Office, and since the years 1995 and 1996 were fertile ground for Web inventions we should start seeing many more Web patents issue soon.
With the Web, futurism has ceased being a luxury: regular visioneering projects are a necessary defense mechanism for anybody who wants to thrive in the network economy where your fundamental business and customer service become automated and thus patentable. Companies that don't claim their stake in the future will wake up in five years and discover that their competitors own all the patents they need to be on the Web.
Toward the end of 1999, some Y2K problems will emerge on the Web. The Web is less susceptible to Year 2000 bugs than most other computer systems:
The Internet runs on modern computers that are "cleaner" than mainframes and embedded systems
Content dominates the Web user experience: plain text and graphics will continue to "run" no matter what the year
The software part of the Web was implemented during the last few years during a time of high visibility for Y2K
Even so, the Web is not immune to Y2K problems:
Integration with legacy systems has been the main way to add functionality to the Web during the last few years - and legacy systems are notorious for their Y2K problems. Few existing companies have rebuilt their entire software base because they started serving customers on the Web
Much Web software was cobbled together in a rush during the frantic years 1995-1997 without any consideration to long-term issues or traditional software engineering principles: who knows what bugs lurk in poorly tested software developed "on Internet time"
Many Web pages use two-digit form fields when users are asked to enter a year
Many URLs and other notations use conventional two-digit representations of the year. I am guilty of this myself, using the file name
981227.html for this Alertbox. Sloppy file names will cause no problems except for the unknown number of cases where a program relies on such notations for computations.
Instead of waiting until December 1999, now is an excellent time to assess your website's Y2K exposure - as well as that of your ISP: if your ISP goes off the net, your company doesn't exist.
See reader comments on this Alertbox.