Michael Gaigg: Über UI/UX Design

3Aug0

Website Analytics I: Browser Support

Posted by Michael Gaigg

Which browser platform should we design for? To most web developers this question is a nightmare and dilemma at the same time, but yet again we face it almost on a daily basis. Just to show you how difficult an answer might be is the disparity between two projects I'm involved in, one for a federal government body that by definition only supports IE6 and another that is expected to be finished in 6 months and therefore was hoping to target IE8. Are these expectations realistic?

Analytics

Let's have a look at some website analytics that were collected over the past year (July 2008 - July 2009), taken from Fanposter.org, a social media website with over 5000 registered users from almost 100 countries and collected through Google Analytics.
Note: While I'm aware that the audience of this website might not be representative for all age groups or countries/regions, it still provides valuable hints about browser usage and trends.

Lesson 1: Internet Explorer on the fall

Internet Explorer is historically bundled with Microsoft Windows Operating Systems. IE6 was released August 27, 2001 just months before WinXP; IE7 was released October 18, 2006 and included with Windows Vista and Win Server 2008; IE8 final was released March 19, 2009 and will be included with Windows 7.

Figure 1: Website Analytics, Browsers Comparison

Figure 1: Website Analytics, Browsers Comparison

The combined total market share for Internet Explorer over the past year was 76.5% but the trend shows a clear decline over the past months from 74.0% in May to 68.8% in June and 53.9% in July.

Will this trend continue to hold true?

A lot will depend on the upcoming release of Windows 7 in October 2009 with Microsoft's power in reaching the people to upgrade their browsers. But almost certainly Microsoft will loose its dominant position of 95% in 2002 and 2003 (IE5 & IE6) or over 75% in the past year to its numerous very capable competitors.

Lesson 2: IE8 on the rise

Figure 2: Website Analytics, Browsers Comparison, Internet Explorer

Figure 2: Website Analytics, Browsers Comparison, Internet Explorer


When comparing the 3 current versions of Internet Explorer out there (IE6, IE7 and IE8 with IE5 and lesser not appearing in the analytics anymore and therefore being disregarded) it seems to be evident that IE8 is on the rise. Though the really surprising observation is that this increase cannot be accounted to users abandoning IE6 but rather to users upgrading their version of IE7. How come? The answer seems to lie in fact that many companies and their IT departments as well as governmental bodies still continue to hold on to and enforce the use of IE6. Can Microsoft overcome these - mostly security-related - concerns through IE8?

Lesson 3: IE6 might not die as soon as we all may wish

Figure 3: Google Trends

Figure 3: Google Trends; The spikes D, E & F indicate the introductions of Beta 1 (March 5, 2008), Beta 2 (August 27, 2008) and Final respectively.


As discussed above IE6 is still a requirement for many users. A quick look at Google Trends reveals that searches for IE6 (Figure 3: blue line) have been pretty stable over the past 4-5 years while searches for IE7 (Figure 3: red line) are declining and searches for IE8 (Figure 3: yellow line) are in the lead. This goes hand in hand with the observations in Lesson 2 above.

In the meantime let's join the numerous battles to kill IE6 like Norwegian Websites Declare War on IE 6 and IE Death March or more seriously at Bring down IE6. Joke or not, IE6 will be around for some more time and therefore needs to be considered if we want it or not.
It won't even help that Digg is discussing to drop support for IE6 or YouTube Will Be Next To Kiss IE6 Support Goodbye. Both sites don't need users that cannot upgrade IE6 e.g. due to limitations at work because exactly these users have limited access restricted by their companies anyway.

Lesson 4: Firefox stays on top

A very interesting observation is the fact that Firefox does a really great job in staying on top of things and keeping their user base up to date. In the last month of the observation period less than 5% of all Firefox users had a version lesser than 3, or in other words, the great majority of Firefox users was using version 3 and higher. And with versions 3.7 and Firefox 4 in the making, a huge base of motivated developers and fascinating plug-ins the future looks bright. Firefox's marketshare seems to climb steadily also, as of July 2009 it showed 32.1% (20.5% average over the past year) with the majority Windows versions and only 0.6% on Mac and 0.2% on Linux. Obviously we all love the high commitment to standards compliancy that makes developing web apps a blast, right?

Firefox 4 Mockup (Win Vista)

Firefox 4 Mockup (Win Vista)

Lesson 5: Chrome & Safari

Google Chrome was released December 11, 2008 and first received as Google's never-ending battle against Microsoft (see graphic by Federico Fieni) but more likely just another smart move to keep users on the web - the faster, the better - or like Ben Parr explains "Revenue = Amount of Time on the Web".

Google vs Microsoft: Chrome

Google vs Microsoft: Chrome

As far as marketshare is concerned, Chrome rose to 2.3% in July according to my analytics. This is still fairly small but not to be underestimated. In terms of testing Google's developer page for Chrome suggests that "if you’ve tested your website with Safari 3.1 then your site should already work well on Google Chrome". But what about Safari?

Safari accounts for 1.3% in the yearly average (with 1.2% on Mac and the rest on Windows). Should we test Safari with a marginal share like this? It really depends on your budget and company structure which might require unjustified additional resources for your QA team. Do the math for yourself: 1% of 10k visitors are 100 potential customers - is it worth the effort?

Lesson 6: Opera is the misunderstood genius

Opera shows an average of 1.1% with an unexpected peak in July 2009 - I need to continue observing where and why that came from. Needless to say, Opera is strong in European countries like Russia and on mobile devices such as mobile phones, smartphones and PDA's (personal digital assistants). It's a shame that Opera is not further ahead, it has been and still is innovator and spearhead in terms of new useful features as well as standards compliancy and implementation.

What are your experiences?

Without giving any decisive conclusions I'd like to hear your experiences or analytics to this subject. For which browser platform do you design for? What do you tell your clients when they ask you this question?

9Nov0

User-Centered Design (UCD) Methods: Comparison and Overview

Posted by Michael Gaigg

This is the first in a series of blogs describing User-centered Design Methods. My goal is to summarize my experience, insights and findings across multiple literature and compile them into easy and quick to digest pieces for you to consume. I want to encourage you to comment your own experiences and give me feedback on why your company applies certain methods differently or not at all or something else altogether.

I personally don't like the term Usability too much, it's an empty buzz word. It means SOMETHING to everybody but isn't scientific enough to be taken serious. It's often interpreted wrongly and purely misunderstood by most. It's kinda like Psychology, we know it is important to understand fundamental human behavior, their problems and remedies, but I wouldn't pay a dime to go to a Psychologist. But who knows, just as Psychology got its scientific relevance and acknowledgment - partly maybe through the 'invention' of the IQ - hopefully Usability rises up to similar levels (Jeff Sauro offers interesting metrics via SUM (Single Usability Metric).

That's why I like the term User-centered Design. It works wonders with Project Managers and the-like, probably because Design is such an important term in their daily work. And when asked about Usability testing I can conveniently point out that this is only one tool of many in my UCD toolbox. But the really important sales trick is to know which UCD method is best used at what time in the project management cycle.

The following chart compares the most common user-centered design methods, outlines their cost and shows when to use them:

Overview of user-centered design methods

Comparison of User-centerd Design (UCD) Methods
Method Cost Output Sample Size When to Use
Competitive Studies Medium Stat. & Non-Stat. 5 Requirements
Focus Groups High Non-Statistical 6-9 Requirements
Field Studies High Non-Statistical 2-3 Requirements
Heuristic Evaluation Low Statistical 2-3 Design
Paper Prototyping Medium Stat. & Non-Stat. 5 Design
Card Sorting High Statistical 15-20 Design
Participatory Design Low Non-Statistical n/a Design
User Testing Medium Stat. & Non-Stat. 5 Design & Evaluation
Surveys Low Statistical 20+ Requirements & Evaluation
Interviews High Non-Statistical 3-5 Requirements & Evaluation
Server Traffic Log Analysis Low Statistical n/a Evaluation
Search Log Analysis Low Statistical n/a Evaluation

Not long ago, after having completed a full project management cycle (requirements, design, implementation and evaluation) the PM proudly announced to perform a Focus Group with his stakeholders. Showing the ready application, he thought, would surely impress them and lead to valuable feedback for the next milestone. This impulse isn't uncommon but has to be fought before it becomes reality. Does he really want to produce MORE and EXPENSIVE requirements? Because that's the output of Focus Groups. Wouldn't he be better off running 2 iterations of User Testing to reveal usability issues or a Survey to receive input from outside the development environment?

Recommendations

  • Requirements:
    • Competitive Studies
    • Interviews
    • Field Studies
  • Design:
    • Heuristic Evaluation
    • Paper Prototyping
    • User Testing
  • Evaluation:
    • Surveys
    • Server Traffic Log Analysis
    • Search Log Analysis
    • User Testing
  • The Usability effort is NOT proportional to the size of the project. Bigger projects spend less percentage on UCD with same effort. Regardless, as a rule of thumb assign 10% of the projects budget for UCD.
  • Faster iterations of prototype design require less testers

References