useit.com -> Recommended Websites -> Spotlights -> 1999 Q2-Q4 Spotlights

Spotlighted Links From April-December 1999

Please note that some of these links may have suffered linkrot. They did work as of the date when they were recommended, but the Web is a highly fluid medium.

December 28, 1999
Take a look at Just Jeans and see if something strikes you as unusual. If you are in North America, Europe, or most of Asia, I bet you will be surprised by the main graphic, but it's a reminder of an important aspect of the Web. Turning to the content, I really don't like the way the most prominent piece of text on the screen is "You're not logged in." (and despite being in color, the text is not even clickable!)
December 27, 1999
Microsoft.com: The history of Microsoft's website, complete with screenshots of 8 different designs. They have re-posted the wonderful photo of the geeky webmaster proudly showing off the server in 1994. I remember usability-testing www.microsoft.com in 1994 and getting many positive user comments on that photo because it showed that some real personality (in contrast to many other corporate sites).
December 24, 1999
The current issue of The Economist has an insightful special section on the Millennium. In contract to most others, they really mean it and cover all the last thousand years. One thought-provoking article actually takes a 2000-year view of the economy and plots average income in Western Europe from Augustus to Tony Blair. During the period from Year 1 to 1750, GNP per person doubled (from $500 to $1,000). Then, during the last 250 years, income increased by a factor 22. In other words, the growth rate changed from 0.04% to 1.2% per year. What caused recent growth to be 30 times faster than the historical norm? And will growth rates be even greater in the Internet economy? This last question is not answered in the article, but the answer could well be yes, given their analysis of the reason for the unprecedented growth during the last 250 years. The article identifies three other periods in human history with great scientific discoveries (the Romans, Islam around year 1000-1200, and China during 800-1400), all of which happened in great empires that were more interested in internal issues than in changing the human condition. Thus, these three periods of discovery did not translate into sustained growth. The lack of a ruling empire in the highly fractioned Europe of the 18th century was the trigger that translated invention into an explosion of income per person. Well, the Internet is even more fractioned and even less of an empire than Europe ever was. So if we accept The Economist's analysis of the history of human income, then the Internet Economy should grow even faster.
December 21, 1999
A collection of 250 personal websites with traffic patterns that follow the Zipf distribution. Just as we have found for the Web at large and for the Web in Australia.
December 20, 1999
Bob Frankston (co-inventor of the spreadsheet) has a great analysis of why WAP is mis-guided. Instead, he is a proponent of IP everywhere. One thing I learned from working with Eric Schmidt is to keep fighting for universal IP and to reject proprietary access. (An aside: some people mistake Bob and me, and if you look at the photo on his home page, you can see why :-)
December 3, 1999
BBC: writing for the radio. Different than writing for the Web, but a good reminder of the importance of writing for the medium.
November 27, 1999
Classic example of clueless merchandising on the Web: What is the difference between the Panasonic S9670 and S7680 S-VHS recorders? Why is the second model twice as expensive? I don't know, so I bought the cheapest one (at 20% off through a dealer recommended by MySimon). It would have served Panasonic right if I had bought a Sharp instead, but what is "Super VHS Expansion Technology": true S-VHS or some kind of emulation or "ready for"? Sharp lost the sale because of poor content.
November 26, 1999
Interesting experiment in special-interest electronic publishing: interviews with 20 leading usability and design experts published as a 136-page electronic document for $3.95. Unfortunately, the special document format only works on Windows, though Mac and Linux versions are promised "real soon now".
November 26, 1999
I bought a 1600-pixels wide flat panel display. Very nice in its ability to view two full-sized Web pages next to each other. The new generation of bigger screens highlights the stupidity of the Maximize button in Windows. This button needs to be redefined to "make the window the largest useful size" as soon as possible. It is harmful to big-screen users to have Maximize mean "make the window take up the entire monitor". This issue will become even more pressing next year when we get the new 4,000 pixels-wide IBM screens.
November 19, 1999
Apple continues its tradition of treating its developers badly. For decades, independent software vendors (ISVs) have complained about the difficulty of working with Apple (whereas Microsoft, Sun, and most other platform vendors cultivate ISVs and treat them like royalty). With the new world of computers as communications devices, content providers are the new developers and you want to encourage them to provide content for your platform. Despite this, Apple is apparently preventing ICP (independent content providers) from getting revenues from the new Sherlock 2 search tool on the Macintosh. Apple simply replaces the website's ads with its own. Of course, advertising doesn't work on the Web, but search ads are one of the few exceptions to the rule (you search when you want to go elsewhere, so a targeted ad based on the user's keywords may have a slight chance of succeeding). Apple has a dying platform on its hands. It should not antagonize developers, whether ISVs or ICPs. More proof that Apple is pursuing a short-term cash cow strategy rather than building a long-term survival strategy for the Mac.
November 11, 1999
The eBay auction of seats in my New Orleans workshop has now closed. It was an interesting experiment, but probably ahead of its time. I got a good deal of email from people complaining that their companies will not allow them to expense items bought at auction. We are all very excited about the new business models supported by the Internet, but the facts of conservative accounting processes will stand in their way for some time (to say nothing of the problems related to tax rules).
November 1, 1999
New data from visit to Sweden: An eye-tracking study of Swedish users has shown that they never have any eye fixations within the advertisements on Web pages. This new study by Iréne Stenfors from Lund University confirms similar studies of American Web users. Nice to know that there are some commonalities across cultures. Stenfors used a combination of eye-tracking and usability observation, leading to an interesting finding: Some users had developed techniques to fixate their eyes on other parts of the screen to avoid being distracted by the blinking advertisements. For example, they moved the cursor in the middle of the screen while waiting for the content to load and kept their gaze focused on tracking the cursor so that they didn't have to look at the ad during the period where it was the only element on the page that had loaded. When users despise something so much that they develop defensive techniques, it's time to stop. (Credit: Computer Sweden, Oct. 22, 1999, p. 8 - article in Swedish only.)
October 28, 1999
Advice for anybody who wants to put a contact telephone number on their web pages: make sure that the call center is staffed 24 hours per day. The Web is truly the place that never sleeps, and you will annoy customers when they see a service number listed but get a recording if they call. If you do close down at "night" (in your time zone, that is), then you can have the server dynamically change the pages to annotate the phone number with a notice stating that it's useless. (I thank Faisal Jawdat from drawing my attention to an ecommerce site that is particularly clueless in this area.)
October 26, 1999
Observation from Zürich: every tram has a URL on it, advertising one Swiss site or another. The Web is definitely happening in Europe.
October 16, 1999
MediaInfo reviews the new 2-column design of the Los Angeles Times' site. I agree that it's annoying that the new design only allows you to utilize half of the window for the content. But since the page layout is liquid, you can expand the window (if you have a big monitor) and get as much room as you want for the articles.
October 9, 1999
Yahoo's revenues for Q3 1999 corresponds to making 0.47 cents per page view (CPM: $4.7), an increase of 6% relative to the previous quarter. Bravo, Yahoo. (Do note that sites with smaller traffic will still not be able to live off advertising at these rates.) Given this good result, I predict that Yahoo will temporarily make as much as 0.5 cents per page view in Q4 because the holiday shopping season usually leads to increased revenues. In other words, Yahoo's predicted revenues for Q4 are $205 million.
September 23, 1999
There is a dearth of strategic thinking about the implications of the international nature of the Web. It counts as leading-edge if a company conducts user tests with foreign users or has a real plan for discontinuing "uplift" pricing schemes and overcoming distributor channel conflicts. Therefore, I was impressed by Clay Shirky's essay about culture as an arbitrage opportunity and the impact of the Web on the Euro and the Pound.
September 22, 1999
Tim Berners-Lee's new book about the history of the Web was published today and is already best-seller #91 at Amazon.com. I haven't read it yet, but I still remember the first time I saw Tim demo the Web in 1991 (one thing I remember thinking is that it was slow - even then). Reviews are mixed so far, but the early days were so exciting that I think they are worth reliving and remembering now that the Web has slowed down (yes, for newcomers who were not into the Web in 1991, 1992, and 1993: those were the days when things really happened - I didn't walk 50 miles to school, but I used the Web in line-mode).
September 19, 1999
The U.K. is changing many phone numbers, changing the city codes for many areas. I estimate the cost of changing a large number of speed-dial buttons in the country to be $20 million (£12M). Only 1/4 of this cost is for the time to actually punch in the new numbers. 3/4 is the cost of time lost trying to figure out how to reprogram the speed dial buttons or to look for the user manual. As usual, usability means big bucks/pounds. (Nick Bull from Cambridge points out that the change has usability benefits as well, in particular for Londoners who will have to dial fewer digits.)
September 10, 1999
A 1996 article in NetscapeWorld about animation turned out to be very prophetic: I was quoted as predicting that over-use of animation in Web advertising would cause users to turn off animation completely. Now, three years later, I installed a browser add-on that does exactly that (WebWasher).
September 9, 1999
Interesting new feature on Microsoft's website: reviews of small-business websites with advice for redesigns. Small companies are often overlooked in the attention lavished on multi-billion dollar Internet start-ups, but small firms have great potential on the Web because of the narrow-cast nature of the medium.
September 9, 1999
The death of the Web browser grows near: Carmen's Headline Viewer is a specialized Internet-access application aimed at browsing headlines from a large number of XML-syndicated sites. No need to manually go to each site in a clunky browser that's really an article viewer, when you can integrated headlines from multiple sites into an interface that's optimized for headline viewing. This new software is another example of the growing trend toward Web clients that are not general-purpose browsers: the Auction Browser was another. Specialized clients can have a user interface that is optimized for a certain class of users and a certain class of tasks - this will always win over a general-purpose UI. Of course, the generic browser will still remain with us for low-frequency tasks. For example, somebody who only buys one thing at eBay would not bother installing a special auction client for that one-time task. So it's not "death of browsers" as software, it's death of browsers as the defining aspect of the Web experience.
September 7, 1999
WebTV is a paragon of usability methodology: regular user tests every week. I only know of two other companies that collect weekly usability data.
September 5, 1999
It is time to take Web maintenance seriously: case study of user being stumped in simple task of researching Melbourne movie tickets.
September 3, 1999
Most poorly written website in a long time: SoftLock. It is impossible to glean any concrete info about the product from the morass of marketese. Besides having a useless website, I also fear that SoftLock has a useless product. The only concrete info I could find was in the customer support section which stated that it may not be possible to read copy-protected documents after running Norton Utilities or other disk utility software. I just can't imagine customers paying top dollars for a report that they may not be able to read if they make any changes to their PC. Copy protection failed for software, and I predict that it will fail for documents as well.
September 1, 1999
The comic strip Waiting for Bob is quite appropriate today.
August 24, 1999
E-commerce case study: trying to buy batteries online: of 45 sites selling batteries, only 3 carried both the two batteries the user wanted (or, possibly, it was not possible to find the batteries on the site - there is no difference between these two situations as far as the user is concerned). Lesson 1: Carry a wide product selection and make it easy to navigate. The cheapest site didn't get the order because (a) they had too high shipping fees and (b) one of the products was so weakly described that the customer could not be sure that it was the one needed. Lesson 2: Good and detailed product descriptions are essential when the customer can't see and touch the products before buying. Content design closed the sale.
August 23, 1999
Usability metric from real life: across six corporate websites, the measured success rate was only 26% when prospective job applicants were asked to find a job opening that was suitable for them and apply for it on the site. Is this because HR sub-sites are worse than other parts of big-company sites? I don't think so: most other pages would score as low if users had to complete a series of steps. A single mistake any step on the way is enough to doom the user to failure.
August 22, 1999
When Bill Gates and I agree on something, there must be something to it: Who Reads DaveNet?.
August 20, 1999
Amazon.com's new purchase circle feature may be a massive privacy violation: they list the books that have been heavy sellers to employees at specific companies. Let's say that I discover that Sybase has bought lots of books about data warehousing. This could be early warning to competitors that Sybase is planning a major push in that area. Maybe not a big surprise, but I could imagine much interesting use of the book sales for business intelligence. Purchase circles are a cool feature, but probably over the line as far as privacy is concerned.
August 19, 1999
Data nugget on the implication of usability for e-commerce sales: After IBM redesigned its site in February 1999, sales increased by 400 percent. Changes included better search that places the product page on top when you search for a product name as well as easy access to specifications of the alternative configurations of the products. "On the old IBM site, tracking down this information was a maddening chore," and so they didn't sell much. (Thanks to Joe Grant for bringing this link to my attention.)
August 18, 1999
Business-process invention disclosure:
Establish a reverse affiliates program where an e-commerce site links to some of the better specialized sites to use their content and recommendations as an enhancement and objective third-party advice for customers. For example, an online bookstore could have a list of "top-10 sites that review and recommend books in the field of X," where X can be anything from major best-selling areas like mystery novels to narrow fields like usability. Amazon's "readers-who-bought-this-book-also-bought" feature could be supplemented with a "these-are-the-main-sites-that-recommended-this-book" feature. Count the incoming referrals from affiliate sites and link back to those that generate the most sales. This can be done automatically: a simple way to add value-added content to every product page. Hint to Barnes and Noble: this would be a cheap way for you to both make your affiliates program more attractive (enticing sites to switch to your side) and to get value-added recommendation content. Of course, this idea works for other industries as well.
August 12, 1999
Two invention disclosures:
When you double-click the Back-button in a Web browser, it should take you straight back to the previous site you were on, by-passing any intermediate pages.
The new personal video recorders like TiVo could eliminate TV commercials completely as a new subscription service: hire somebody to watch each major channel and to push a button when commercials come on and are done. Publish this data stream on a subscription-only website. Have a separate company sell a program that deletes all frames within specified time intervals. Have the customer hook the software to read the data feed. Presto: smooth uninterrupted programming when watching a TiVo recording.
August 10, 1999
In preparing for the forthcoming holiday shopping season, e-commerce sites need to worry about what products and services are giveable things. It will be a challenge to find ways of making online services feel like good presents. CDNow has a great idea for a giveable service: personalized music CDs which you can order over the Web and decide what sound tracks go where. Presumably, you can also design a personalized cover and maybe even write your own liner notes. Should make a hot gift item, at least for people who agree with the slogan "it's the thought that counts" and are willing to spend some thought designing their own individualized gift. The Web should support an explosion in such custom-designed gifts that allow the givers to express their own creativity and thoughtfulness as the same time as they (hopefully) match the recipient's personality better than any mass-produced item.
August 5, 1999
Password problems account for 20-50% of all calls to help desks, costing about a million dollars per year for a typical mid-sized company. About time we start taking the human factors of security seriously. Passwords don't work: except for certain circus performers, nobody can remember a large number of random strings. And yet, how many security groups include a usability expert? (New York Times article: access requires free registration.).
August 2, 1999
Eric Davis, an Information Architect with Resource Marketing, recently reported on a usability test of shopping cart terminology. Actually, the draft design he tested featured the term "Shopping Sled" since the site (selling a winter sports products) was interested in standing out and avoiding standard terminology. Result: "50% of users did not understand The Sled concept. The other 50% said that they figured out what it meant because it was in the same location as a cart would be. They knew that you had to add to something, and the only something that made any kind of sense was the Sled." Lesson: Do not try to be smart and use new terms when we have good words available that users already know. Remember Jakob's Law of the Web User Experience: Users spend most of their time on other sites. So anything that's a convention and used on the majority of other sites will be burned into the users' brain and you can only deviate from it on pain of major usability problems.
August 2, 1999
Bruce "Tog" Tognazzini analyzes the strengths and weaknesses of interface paradigms, including command line, early GUIs, modern GUIs, and contemporary Web browsers. As we all know, browsers lose. I particularly like the reference to Web browsing being an order of magnitude slower than a 1978 Apple II. Tog then moves to discussing when you should implement applications in browsers (rarely) and what should be done about it.
July 16, 1999
A new task-specific browser has been introduced that is optimized for accessing auction information on eBay. I expect we will see many more such specialized user interfaces to supplement the generic browser. IE or Netscape are fine for reading articles (even though they need better navigation features) but clunky as platforms for real applications where users need to do something such as monitoring and managing a set of real-time resources (auctions, in this case). It is unfortunate that the new specialized UI is targeted at a single site (eBay): most likely it will be difficult to use it for managing auctions across multiple auction sites, even though cross-site integration would be a highly useful feature of a task-specific browser. For additional analysis of the need for specialized Web software, see the Alertbox for September 1998: Does Internet = Web?
July 13, 1999
A new initiative to define a standard for e-commerce will codify best practices and set the bar for treating users decently online. A draft document is available right now and the sponsors will soon open the site for public comments and votes on the individual rules in the standard. See Jesse Berst's article about the standard for a short summary of the project.
July 8, 1999
Analysis of the impact of language on the Internet yields a rare Deep Insight into the nature of the network economy where topology and trading zones are being defined by shared language and not by geographical proximity or political rules. English and other human languages are becoming more important for defining the Internet economy than programming languages. The Japanese vision of smooth natural language translation is not going to happen for 30 years, so we need to understand the impact of human languages on the Web. No, not everybody will speak English. Yes, right now "the English language [is] playing the role of the Royal Navy" in building the Internet Empire. It is an interesting question whether France and Sweden will contribute to the world economy as anything except a fringe ethnographic curiosity in the future. Metcalfe's Law probably applies to human languages as well: the value of a language network's economy may be the square of its components. So the square of Swedish is Sweden. But the square of (USA + UK + Australia + Singapore + India + South Africa, etc.) could be be world dominance. Hollywood was just the beginning. We don't know, but it's surely worth more thought. Especially if you speak nothing but French which could be the 21st Century's equivalent of living in a straw hut. (Before I get flamed by L'Académie française, let me say that I am not advocating this, simply pointing it out as an issue for study. In fact, I think it is dangerous for a company to base its Internet strategy on an assumption that English will continue to rule.)
July 8, 1999
A new Microsoft press release about ClearType seems to position it for the narrow market of electronic books. It would be a disgrace if Microsoft were to sit on an invention that can improve reading productivity by 25% for all users. Maybe we need to start a campaign to get ClearType into Windows 2000 since it will cost the world economy about half a trillion dollars if they delay ClearType until Windows 2002.
July 7, 1999
The new NEC study of the size of the Web estimated 800 million pages in February 1999. Thus, the current growth rate of the Web is 163% per year in terms of pages; in comparison the growth in number of sites is "only" 145% per year. This difference makes sense since the old sites also keep adding pages. Using the growth rate to predict the current size of the Web gives 1.2 billion pages right now and 2.1 billion in January (when there will be about 10.4 million sites). The study is widely cited for finding that search engines don't index all of the pages, but this focus on search recall is misguided: the larger the Web gets, the less important it becomes whether a search can retrieve all info about the query since nobody has time to look at that much data. It is more important whether the search finds the best info, and that's where many of the search engines are failing: they surely don't improve the quality of their search algorithms by 163% per year. I don't think the main Internet search engines get better at more than about 10% per year.
July 7, 1999
Yahoo's quarterly earnings for the second quarter of 1999 again come out to 0.4 cents per page view. The same as last quarter and the same as last year (except for the December quarter where they succeeded in "monetizing" their traffic at 0.5 cents per page view; probably because of the strong holiday shopping on the Web). It is remarkable that Yahoo has been able to hold a steady monetizing quotient despite the generally declining value of Web advertising. Good for them, I guess. It is also nice to see that some aspects of the Internet economy are becoming firm and predictable. I would also claim that this data further supports my claim that advertising cannot be the foundation of the Internet economy since only the very largest sites (like Yahoo, of course) can live with a monetizing quotient of 0.4 cents per page view. MQ=0.4c (not quite the law of relativity, but maybe the law of Yahoo revenues).
June 27, 1999
Iconocast reports that online marketers are shifting their advertising budgets to real-world media: currently e-businesses only spend 44% of their advertising budgets outside the Web, but survey respondents reported plans to boost this to 52%. These people are starting to realize that advertising doesn't work on the Web, so they are spending more of their money where it will do more good.
June 27, 1999
In its review of the new Mercedes S-class, The New York Times complains that "instead of logical buttons and knobs, Mercedes has come up with page upon page of on-screen menus. Simple tasks like lowering the bass or rewinding a tape are maddening." And the phone system annoyed the reviewer a few times too often with an error message saying "that function is not possible." Maybe the developers of this $80,000 car could not afford a usability engineer on the project team.
June 21, 1999
Who owns my home, my office, and my tools? There is an unfortunate tendency for outsiders to buy preferred treatment in our tools. The latest two examples were announced today: It's time for consumers to say stop and regain ownership of their own home and office. How about a TV set that you turn on to watch ABC, but it doesn't allow you to do so without first watching a minute of MSNBC? That's the equivalent of getting Yahoo into the PC chips. Or a telephone that plays commercials while you are dialing your mother's number? That's the equivalent of Lycos getting a button on Lotus Notes. Soon we will have refrigerators that play the Coke jingle every time you open the door. Cute as an optional add-on for fans, but an intrusion if made standard on every fridge.
In addition to violating the "my home is my castle" principle, these schemes hurt productivity in the workplace. Instead of having their PCs dedicated to optimal productivity (with the intranet home page as the starting point), the machines will spout distractions attempting to get employees to go to websites unrelated to their jobs. MIS departments will be wise to question whether to allow company staff to use this new release of Notes or PCs with contaminated BIOS chips. Doing so could easily cost millions of dollars per year in lost productivity.
June 19, 1999
True story. Note enclosed in recent mailing from US Air: "Due to recent system enhancements, your account number has been changed as noted on the front of the Award Claim Form." Such blatant disregard for customers is what gives computers a bad name. Of course there is no need for a "system enhancement" to interfere with customers if you have minimal respect for their time in planning the new system architecture. Sure, it might have required a few extra hours of programming for the new system to handle old account numbers, but isn't it better for the company to suffer than for consumers to suffer? Any time you get sub-standard customer service with the excuse being "because of the computer," you know that you are dealing with a company that still doesn't get it with respect to the coming Internet economy.
June 11, 1999
The Association of National Advertisers (ANA) has released a survey of Web marketing by huge companies (that spend an average of $380 M on all marketing per year). Two main findings:
  1. Maintenance budgets are 72% of the cost to develop a site in the first place (I have always recommended between 50% and 100%):
    • Average cost of developing a new site was $252,500
    • Average cost of maintaining a site was $182,000 per year
    • E-commerce sites are more expensive than info-only sites, with average development costs of $307,100 (22% more) and average maintenance costs of $275,000 (51% more)
  2. Less spending on Web advertising (which doesn't work; people who run $380M marketing departments are pretty smart and are rapidly figuring this out):
    • Percentage of these big companies that advertise on the Web declined from 68% in 1998 to 61% in 1999
    • Spending on Web advertising for those companies that still advertise has fallen from $714,000 to $649,000 (about 0.2% of these companies' total marketing budgets)
June 8, 1999
Today's DaveNet about the Office 2000 rollout asks a good question: "What is Office? One or two sentences please." I have two answers:
June 6, 1999
A much-reported study shows that registered users spend much more time on portal sites than non-registered users. However, the data doesn't prove that registration in itself is the cause of this added use. An equally likely hypothesis is that those users who are already heavy users of a site are the ones with the most motivation to register.
Similarly, the fact that the registered users concentrate their use at a single site as opposed to going to multiple "portals" could maybe be due to the registration, or it could be because the people who register at a single site are those who already are pre-disposed to liking it more than other sites. In contrast, people who like to use a multiplicity of sites may not bother registering because they would have to do so multiple times with current technology (from a usability perspective, it would be better with a single, Web-wide, registration).
Without looking at a control group with similar pre-registration behavior, there is simply no way of knowing the causality underlying the observations. The alternative explanations probably account for much of the observed effect and the registration itself probably does cause some amount of increased use. But we can't say how much.
June 5, 1999
Lack of coordination strikes again: Today, McAfee gives its customers three different versions of its virus definition file, depending on the method used for the download: It is particularly embarrassing that paying customers get an obsolete virus definition file when they download it as part of a Web-based purchase. The probable explanation is that the downloadable software is treated as an old-fashioned "manufactured product" that is kept frozen between each "golden master." Concepts that make no sense on the Web: when one of the modules in the product is updated, then the "master" should be updated immediately and not treated like a CD-ROM (which you only change when you run out of the previous print run).
June 1, 1999
The UI Program Manager for Internet Explorer expresses his frustrations about the difficulty of improving Web usability in the June issue of AskTog. I have been pretty harsh on IE at times (complaining about the poor design of Netscape as well), but I admit that it is hard for the IE team to get improvements through when websites don't want to reauthor for usability. For example, they tried in vain to push a much-needed feature for getting "websites to indicate to the browser how documents were related to each other." The current history bar in IE5 is a step in the right direction, but as noted by the designer, "sites can spoil [it] by poorly titling their pages" - clueless microcontent is legion on the Web. So on the one hand I will grant Microsoft that the poor usability of the Web is not their fault. On the other hand, they could do more: for example build better authoring tools that promoted usable design, was integrated with a set of recommended design standards, and encouraged use of the good features of IE while generating high-quality code that would work for users who still have non-IE browsers.
May 31, 1999
On average, Web users tell 12 other people about their online shopping experiences, confirming the importance of the social interface to the Web. In comparison, people tell 9 others about their movie-going experiences and word-of-mouth is known to be an extremely powerful mechanism for deciding which films become popular. Having 33% more word-of-mouth for the Web is additional proof of the user-driven nature of the Web. The user interface must be great for every single visitor: if a user likes the site, 12 others will hear that it is worth visiting; if the user had a bad experience (for example due to using an old browser that crashes with a JavaScript error), 12 others will be told to stay away.
May 29, 1999
I dare you to find the best CD with music by Mendelssohn that includes his Violin Concerto in E Minor on Amazon.com. It's virtually impossible (unless you accept a mixed CD with works by other composers) and proof that Amazon is not going to rule all parts of e-commerce: buying classical music requires a very different user interface than buying books. In general, specialized services can offer a better user experience than generic services. CDNow is slightly better than Amazon, but not good enough; I don't think anybody has cracked the problem of an acceptable UI for buying classical music. The basic rule of ecommerce remains: make it really easy to buy - and you will sell!
May 26, 1999
USA Today should be much applauded for using different headlines in their printed newspaper and on their website. The requirements for online headlines are very different than in print. For example, a story on two headline writers for portals has a print headline of "Twosome tells wired world what's news" (cute, but useless as a hypertext link) and "Bringing news to the wired world" as the link in the list of technology articles (see entry for May 19). The Web link is not a perfect headline for the article since it's too generic and doesn't hit at the issue of headlines or compressed writing. There is an odd dual goal of attracting readers to the article while still protecting users from clicking on anything they won't be interested in (every time you trick people to click to something they don't want, you erode your brand and reduce the chance that they will click again).
May 24, 1999
John C. Dvorak has an absolutely correct analysis of why it is better to be a focused site than a bland portal that tries to do everything. The vertical portal opportunity is screaming at us and nobody is picking it up (no, good as MiningCo/About.com is, it's nowhere near what we need). A "just-add-domain-expertise" platform provider could be worth billions. Any VCs listening?
May 23, 1999
I have finally come across a good book about writing for the Web, appropriately titled Writing for the Web, by Crawford Kilian.
May 22, 1999
AltaVista must have done something to improve the quality of its search results drastically. I don't know what and I haven't seen any press releases from them, but over the last week they have become almost as good as Google.
May 15, 1999
I am receiving a rapidly growing flow of press releases and PR email. Most of these are wasted since I rarely write about products but focus on the big picture and long-term trends. But it's interesting that Web columnists are now seen as an important target for PR. Of course, this is a correct insight on the part of Internet-savvy PR folks since my articles usually attract about 100,000 readers. On the other hand, most of these readers access the columns a year or more after they were written, so getting "ink" (or rather, "pixels") in my writings would not benefit most traditional, short-sighted companies. The Web is a new publishing paradigm with eternal archives (at least on well-designed sites) so it may require new thinking among PR strategists to learn how to best deal with persistent publishing as opposed to ephemeral publishing as known in old media (and poorly-run sites).
April 26, 1999
The Yale Web Style Guide is now available as a hardcopy book. I tend to prefer more principle-based or methodology-based books, but there is also something to be said for a long list of detailed design rules. And the online version of the Yale styleguide has for sure proven to be a classic over the years.
April 25, 1999
Bruce Tognazzini explains why he likes writing long Web pages. The good Tog ought to recognize that only a brilliant writer like himself can keep users scrolling to the bitter end. The average site is cursed with extremely impatient users who want to get in and out and get answers or buy products fast. Paradoxically, the average site probably has below-average writing, since most commercial sites use repurposed print writing filled with "marketese" which backfires in terms of lowered trust and consumer skepticism. So, read Tog, but don't listen to him when it comes to page length.
April 22, 1999
New evidence supports my 1997 theory that the size of websites follows a Zipf distribution: Data on the distribution of traffic in Australia shows that The first statistic is a little off from the theory, but the second statistic is exactly right. In general, if two sites are rank-ordered N and M for traffic, then the ratio between their traffic is N/M, so comparing site #100 with #500 gives a ratio of 100/500 = 1/5 = 20% (exactly as measured). Interesting to know that the theory holds up not just for the Internet as a whole but also when considering a subset of the net (here, Australian users accessing Australian sites). (I thank Eric Scheid from Ironclad Networks for bringing this data to my attention.)
April 16, 1999
AltaVista has posted its proposed design for taking payment for search results. Considerably better than some commentators (including myself) had feared: the untrusted area is clearly marked as "Paid Placement." Such honesty is the best policy for creating long-term trust and value on the Web. The new AltaVista design still degrades the service, however. There are now four paid links that distract the user from the actual search results:
  1. The traditional banner ad (presumably still sold on a keyword basis)
  2. Whoever bought the keyword at RealNames
  3. Books from Amazon.com (many users quickly learn not to click this link since it rarely leads to the best books about the topic)
  4. The new "paid placement" link
Clutter, clutter, clutter. On the sample screen, the search results are allocated about 1/3 of the pixels - but a user with a small screen would not see any search results above the fold. One wonders how long a search engine that doesn't show search results can survive.
April 16, 1999
Update to earlier posting: I assumed that the difficulties in using the Canadian Imperial Bank of Commerce's home page were due to response time delays. This is probably wrong. Instead, the main usability problem is that the active area only cover a small part of each of the big buttons. You cannot have a button that looks like a single interaction element but behaves differently, depending on what invisible area the cursor is over.
April 15, 1999
vBay, a parody of eBay.
April 7, 1999
During Q1, Yahoo made 0.4 cents per page view. Down slightly from 0.5 cents for Q4 of last year as reported in my Spotlight for January 13. Since the new number is the same as my estimate for earlier in 1998, it may be that 0.4 cents is Yahoo's natural level and that the temporary increase in Q4 was due to the holiday season.
April 7, 1999
Xerox launches a redesign that is claimed to make it easier to compare products. It is indeed great to have links from individual product pages to pages for similar products, though the cross-references don't specify how these other products differ from the current one. It would have been better to order the products along meaningful dimensions and suggest alternatives in terms that make sense to users (e.g.: faster vs. slower, higher-resolution vs. lower-resolution, BW instead of color, etc.). Amazon.com is restricted to viewing its products as an unstructured soup ("other relevant books"), but Xerox should know why it has different product offerings. Despite some design improvements, I challenge you to discover why the Xerox WorkCentre XD100 costs $20 more than the XD102. It's there, but very well hidden and obscured by a content error. The product summary for both models is "Everything you want in a copier, including a printer" which is essentially meaningless and evidence that poor mini-content often dooms the usability of otherwise decent site design. One more usability blooper in the new design: links don't change color after you have visited the destination page - sure way to make users feel more lost. I wouldn't say that Xerox completely wasted the $5M they spent on the redesign because the site does seem better, but it's amazingly poor for the high prize tag.
April 7, 1999
Usability improvements in the redesign of Salon: Missing: subheads, bulleted lists and other aspects of writing for scannability (Alertbox Oct. 1997); liquid layout that adapts to the user's preferred window size (Alertbox March 1997).
April 7, 1999
Update to Spotlight for March 31, 1999: Amazon.com recants and removes the irrelevant auction listings from the middle of their book descriptions. The original design was uncommonly bad at polluting the user experience, but at least it lasted less than a week.
April 4, 1999
A few days after the Los Angeles Times quoted my critique of the website for Bill Gates' new book, the site has been made (reasonably) accessible for users with disabilities. The power of the press. (The link points to syndicated version of the article on the Denver Post site since the LA Times removes articles from their own site after a week.)
April 3, 1999
Forbes has a great article on the penetration of the Internet in Western Europe. The percentage of disposable income one needs to spend to get online from home is 2% in the U.S. and 4% averaged across Europe. In other words, European pocketbooks are hit twice as hard, making it no surprise that there are still more Americans online. Internet costs and available income both vary significantly across countries. Based on the numbers given by Forbes, I have calculated the relative home-budget impact of Internet use in each country, using the United States as a baseline at 100%. Low-cost leaders are Finland 101%, Denmark 122%, and Norway 131%. Somewhat surprisingly Italy at 157% and Belgium at 169% beat Sweden at 171% and The Netherlands at 180%: probably because the survey looks at after-tax income. High-cost countries are Portugal 405%, Greece 371%, Spain 285%, Ireland 263%, and Germany 253%.

Earlier Months in 1999

Other Years