Joel Spolsky (www.joelonsoftware.com), founder of Fog Creek Software and software best-practices teacher, notes in his Joel Test that "hallway usability testing", grabbing average, nontechnical people at random from the hallway and having them test or review your work, is a simple but effective way to improve your software. Since our customers are not themselves software companies we work to educate them about the importance of software testing – we test our software before we give it to them but once they change the user interface they need to test it again for usability.
Steve Krug, author of, "Don't Make Me Think, A Common Sense Approach to Web Usability", my favorite book on the subject of website usability, has a couple of quick points that I think are important for testing – because most people don't like to do testing:
1. If you want a great website you have to test.
We've already established that you need to test before your customers do. Customers are the worst testers because they just assume that your site will work the way they expect it to and they are not forgiving if it doesn't. This means that if you paid money to get them to your site and so when they are unhappy you not only lose that direct marketing money but also any possible future revenue from them as a loyal returning customers, not to mention their network of friends – "Their site was buggy or hard to use", won't draw their friends to your site. Also, visitors are busy and rarely will any of them drop you a note of any kind to let you know that there is an issue with your site. We have seen sites with Customer Experience Management (CEM) links and buttons – the "Report a problem with our site" – but our own experience on our own website and discussions with others that have employed CEM themselves is that it doesn’t work as you expect it to work in catching issues because people are just too busy and CEM is not for testing for bugs, it's for providing another channel for customers to communicate with you.
2. Testing with a single tester is 100% better than testing with none.
The math is simple, if one person tests then you are 100% better off than having zero people test your site. A few caveats though, not everyone is going to be a good tester and testing is not a one-time event. Testing needs to be done every time you make a change to the way your site functions.
3. Testing one user early is better than testing 50 users near the end.
Testing at the beginning is cheaper and better because once you build something there is all the extra effort that you must expend that you wouldn’t have had to have expended if you had tested earlier. Imagine that you are building a house and you decide after the house that you want another bathroom up stairs. If your house is finished then you have to wreck the walls, floor and plumbing to add the extra bathroom but if you had tested earlier and discovered that an extra bathroom was needed you could have added the bathroom at the start and not had to suffer the cost of building once and then wrecking and rebuilding a second time.
How to test your website on the cheap:
Steve Krug calls this "Lost Our Lease Testing":
1. Who are these mythical and magical testers? A tester is any reasonably patient human being, ideally someone who uses the web and has some experience buying online. Try to find users who reflect your target audience but don't get too caught up on this point because remember that having someone is better than having no one. Also, if you get more than two testers you are more likely to notice the difference between folks who are good testers and those that are not – take it from me the difference is huge and finding a really good tester is critical and hard – you don't need the best (yet) you just need someone who is acceptable and will get you started.
2. How many testers? Ideally 3 or 4. You want more than one so you get at least two opinions and when you get three or four then you should start to see patterns of problems. A single person is still better than zero though and two is twice as many people... you get the point.
3. Where to test: In person at your office or conference room where you and your team can observe them in person and possibly even video tape them. From personal experience, avoid, where possible, having them do the testing remotely and phone it in. If this is all you can get then do it but ideally you want to see them in person because you will be amazed at the level of detail that you will catch just watching someone try and navigate what you thought was a completely intuitive website! My favorite moment is when they fill in the form "wrong" and it causes an issue with the application – what were they thinking is usually the comment, to which I respond, "fix it".
4. Budget: $50/person or lunch or dinner depending on whether these are your friends or not.
5. How long should they test? No more than 1 hour at a time – think of how long your average shopper on your site spends on each visit, one hour is a long time to get through all the basics and some of the more advanced stuff. Ideally you want them to perform at least the basic tasks of buying a product, searching for a product, browsing for a product, and checking out. If you have time get them to also try and cancel an order, check order status, and even find answers to their questions. I would suggest that you record all the different ways that these folks use your site so that each tester is given the same set of tasks and you can compare them. Ideally though you want to start by just telling them to buy a couple of products and tell them to do it the way they always shop and see what and how they do it – I guarantee you will be surprised.
The take-away:
Test your website because your customers won't and it's cheaper, faster and easier than letting your potential customers leave because of something you could have found and fixes for the price of some pizza and beverages.
Posts
