Through your Visitor's eyes

Originally published in "Managing High Impact, No Nonsense web pages for the Busy Executive" 1998.

Last Edited 2005-09-06

Envisaging your web site user involves picturing both what he thinks and what he actually sees. These things can be incorporated into a persona. (Here called Stuart.) Looking through Stuart's eye's helps clarify your thinking and saves time, when designing a web site. A lot of what he sees is determined by his operating system and browsers.

By basing Stuart's preferences on realistic data, the web team:

• Can create a design more easily
• Has more confidence in that design, and spends less effort worrying
• Has a better chance of pleasing real visitors to the site.

Two aspects of how Stuart is connected to the web are his operating system and browsers. These are illustrated, below, with graphed data over about a decade. Such graphs can be used to check how realistic Stuart is. They also illustrate how the situation changes over time.

Operating system

Stuart currently use the XP versions of the Windows operating system (OS). He adopted it about half way through it's rise in popularity (mid 2002). (He is reluctant to spend the time converting to a new operating system version and may delay the next change-over, till a similar point.) He keeps the operating system up to date with the latest service packs and patches.

Notes for the development team and testing panel.

• The exclusive use of Windows frees the team from the need to consider browsers on other OS's.
• Development and testing should be undertaken on a Windows computer to ensure the site is compatible. In other words the test panel should have access to such computers.
• The exact Windows version of the computers is not critical provided that equivalent browser versions are available. (Keeping the system up to date is not crucial for the team, though it is advisable.)
• When a new OS version is released, there is time to get ready for it. This while it is still gaining popularity. Timing can be planned from graphs like these.
• If available, a Mac test is a good idea, even though it's not required by the Persona.
• Stuart reflects the same OS preferences as 80 to 90% of site visitors over the last decade.

Browser

Stuart uses both Internet Explorer (IE) and a Firefox (Gecko) browser. In both cases he has the most recent version of the browser. (It is possible that his Gecko browser is a few months old.) He uses IE more often.

Notes for the team.

• Stuart's browsers reflect the population shown in the graphed data.
• Two browsers, with a single version of each is relatively easy to test. The test panel should ideally include people with both browsers. In times of transition, when one browser version is replacing another, Stuart will likely use both versions. That means both versions should be tested. Installation, of two versions, is easy for Netscape / Gecko browsers like Firefox. For IE it is not as easy, although it is possible. For that reason testing on dual IE versions might only be undertaken by some panel members.
• Historically browsers change more rapidly than operating systems. The exception on the graph is where the Netscape browser had a bad version.
• Even though the code base changed, Netscape and Gecko browsers are considered to be from a single family.
• This simple browser pattern for Stuart (2 browsers), appears to accommodate over 85% of site visitors so is a fairly safe bet.

What screen is he using

Stuart mostly uses a 1280x1024 screen size, sometimes 800x600.

Notes about screen characteristics.

• This dictates what sizes you should use during testing. How that happens depends on your situation. One approach is for the test panel to use their normal machines. You can capture those details during testing, inspect the machines or only keep notes where it's easy. In that case (panel uses what it wants to), there should be tests across the different screen sizes on at least one machine.
• Though a browser safe palette is still a good idea, it is not essential for this persona.

Javascript and Cookies

Stuart has both JavaScript and Cookies switched on always. (He also has a cookie viewer and knows how to switch on JavaScript debugging, though he doesn't do these things very often.)

Some notes.

• For a simple site you may be able to avoid using JavaScript and Cookies. In my experience however, features that require them are always on the table. That makes it wise to plan for them up front.
• There is debate about whether users have these switched on. Surveys suggest that JavaScript is usually on for more than 90% of visitors. Cookies are probably on for over 95% of visitors. Considering how unpleasant the web is without them, I think it reasonable for Stuart to always use them.
• You can plan for a panic where either or both are switched off. Consider what it will do for your site. You can (and should) minimise the impact by design, where that's sensible. Where you depend on these features, there's no point in worrying about it.

Summary

Stuart uses Windows XP and occasionally Windows 2003. When a new operating system is released he will generally use it within about 18 months. He mostly uses Internet Explorer. Sometimes he uses Firefox. For both browsers he quickly adopts the latest version (and patches).

He generally uses a screen resolution of 1280x1024, with an occasional use of 800x600. In both cases he uses a colour depth of 24 bits or greater. JavaScript is always turned on, as are cookies.

Discussion

This profile is deliberately simple. Less browsers, less operating systems means less work. The fact that Stuart applies software updates quickly, helps save time and effort in the development team. The test panel should see the web site in one of the ways he does.

Even though the persona is simple it is based on real life data. This same data may give an indication of your Internet population, it may not. If you know your audience is specialised, like Firefox enthusiasts, or Mac using graphic artists your persona will be different from the Stuart described here.

Stuart is based on aggregating all visitors to a site. This won't be right in every case. If you have doubts, it's worth checking the profile of your real customers, and contrasting that with all site visitors. There could be a significant difference. You probably want your persona to reflect the customers more than it does the all comers profile. (While this is a real possibility, but I have not encountered it.)

The testing that derives from this persona is not the whole story. Irrespective of Stuart's persona, the web design is intended to work well for all users. There are techniques to ensure that. The persona defines the main thrust of development effort. The technical team will run deeper checks, even if the test panel does not.

In the past minor browser version changes have removed features. If you were dependent on such features you could be in trouble! While I see no need to run exhaustive checks for minor version changes, major versions deserve to be re-checked. Before the new browser version is heavily used there is an opportunity to make sure the site works well and adapt if needed. (Don't blame the web developers if the browser changes!!) Even if there is a major problem, there is time to react. For a time few people are using the browser, put another way Stuart only uses the new version for one visit in a hundred. If you react while usage is low little damage will be done.

Further Information

My impression is that there is less good hard information about these details than there was six or seven years ago. Here are some sources that might be as useful to you, as they have been to me.

• The University of Illinois, publishes records of browser statistics going back to 1996. Very useful to get a cross section of web history.
• The W3Schools publish data about a variety of issues like browsers, operating systems and screen resolution.