Don't Make Me Think, Revisited: A Common Sense Approach to Web Usability

by Steve Krug

  It gives you what you need. Watching three participants, you’ll identify enough problems to keep you busy fixing things for the next month.

  It frees you from deciding when to test. You should pick a day of the month—like the third Thursday—and make that your designated testing day.

  This is much better than basing your test schedule on milestones and deliverables (“We’ll test when the beta’s ready to release”) because schedules often slip and testing slips along with them. Don’t worry, there will always be something you can test each month.

  It makes it more likely that people will attend. Doing it all in a morning on a predictable schedule greatly increases the chances that team members will make time to come and watch at least some of the sessions, which is highly desirable.

  How many users do you need?

  I think the ideal number of participants for each round of do-it-yourself testing is three.

  Some people will complain that three aren’t enough. They’ll say that it’s too small a sample to prove anything and that it won’t uncover all of the problems. Both of these are true but they just don’t matter, and here’s why:

  The purpose of this kind of testing isn’t to prove anything. Proving things requires quantitative testing, with a large sample size, a clearly defined and rigorously followed test protocol, and lots of data gathering and analysis.

  Do-it-yourself tests are a qualitative method whose purpose is to improve what you’re building by identifying and fixing usability problems. The process isn’t rigorous at all: You give them tasks to do, you observe, and you learn. The result is actionable insights, not proof.

  You don’t need to find all of the problems. In fact, you’ll never find all of the problems in anything you test. And it wouldn’t help if you did, because of this fact:

  You can find more problems in half a day than you can fix in a month.

  You’ll always find more problems than you have the resources to fix, so it’s very important that you focus on fixing the most serious ones first. And three users are very likely to encounter many of the most significant problems related to the tasks that you’re testing.

  Problems you can find with just a few test participants

  Also, you’re going to be doing another round each month. It’s much more important to do more rounds of testing than to wring everything you can out of each round.

  How do you choose the participants?

  When people decide to test, they often spend a lot of time trying to recruit users who they think will precisely reflect their target audience—for instance, “male accountants between the ages of 25 and 30 with one to three years of computer experience who have recently purchased expensive shoes.”

  It’s good to do your testing with participants who are like the people who will use your site, but the truth is that recruiting people who are from your target audience isn’t quite as important as it may seem. For many sites, you can do a lot of your testing with almost anybody. And if you’re just starting to do testing, your site probably has a number of usability flaws that will cause real problems for almost anyone you recruit.

  Recruiting people who fit a narrow profile usually requires more work (to find them) and often more money (for their stipend). If you have plenty of time to spend on recruiting or you can afford to hire someone to do it for you, then by all means be as specific as you want. But if finding the ideal users means you’re going to do less testing, I recommend a different approach:

  RECRUIT LOOSELY AND GRADE ON A CURVE

  In other words, try to find users who reflect your audience, but don’t get hung up about it. Instead, loosen up your requirements and then make allowances for the differences between your participants and your audience. When somebody has a problem, ask yourself “Would our users have that problem, or was it only a problem because they didn’t know what our users know?”

  If using your site requires specific domain knowledge (e.g., a currency exchange site for money management professionals), then you’ll need to recruit some people with that knowledge. But they don’t all have to have it, since many of the most serious usability problems are things that anybody will encounter.

  In fact, I’m in favor of always using some participants who aren’t from your target audience, for three reasons:

  It’s usually not a good idea to design a site so that only your target audience can use it. Domain knowledge is a tricky thing, and if you design a site for money managers using terminology that you think all money managers will understand, what you’ll discover is that a small but not insignificant number of them won’t know what you’re talking about. And in most cases, you need to be supporting novices as well as experts anyway.

  We’re all beginners under the skin. Scratch an expert and you’ll often find someone who’s muddling through—just at a higher level.

  Experts are rarely insulted by something that is clear enough for beginners. Everybody appreciates clarity. (True clarity, that is, and not just something that’s been “dumbed down.”) If “almost anybody” can use it, your experts will be able to use it, too.

  How do you find participants?

  There are many places and ways to recruit test participants, like user groups, trade shows, Craigslist, Facebook, Twitter, customer forums, a pop-up on your site, or even asking friends and neighbors.

  If you’re going to do your own recruiting, I recommend that you download the Nielsen Norman Group’s free 147-page report How to Recruit Participants for Usability Studies.4 You don’t have to read it all, but it’s an excellent source of advice.

  4 ...at nngroup.com/reports/tips/recruiting. It’s from 2003, but if you factor in 20% inflation for the dollar amounts, it’s all still valid. And did I mention that it’s free?

  Typical participant incentives for a one-hour test session range from $50 to $100 for “average” Web users to several hundred dollars for busy, highly paid professionals, like cardiologists for instance.

  I like to offer people a little more than the going rate, since it makes it clear that I value their time and improves the chances that they’ll show up. Remember that even if the session is only an hour, people usually have to spend another hour traveling.

  Where do you test?

  To conduct the test, you need a quiet space where you won’t be interrupted (usually either an office or a conference room) with a table or desk and two chairs. And you’ll need a computer with Internet access, a mouse, a keyboard, and a microphone.

  You’ll be using screen sharing software (like GoToMeeting or WebEx) to allow the team members, stakeholders, and anyone else who’s interested to observe the tests from another room.

  You should also run screen recording software (like Camtasia from Techsmith) to capture a record of what happens on the screen and what the facilitator and the participant say. You may never refer to it, but it’s good to have in case you want to check something or use a few brief clips as part of a presentation.

  Who should do the testing?

  The person who sits with the participant and leads them through the test is called the facilitator. Almost anyone can facilitate a usability test; all it really takes is the courage to try it, and with a little practice, most people can get quite good at it.

  I’m assuming that you’re going to facilitate the tests yourself, but if you’re not, try to choose someone who tends to be patient, calm, empathetic, and a good listener. Don’t choose someone whom you would describe as “definitely not a people person” or “the office crank.”

  Other than keeping the participants comfortable and focused on doing the tasks, the facilitator’s main job is to encourage them to think out loud as much as possible. The combination of watching what the participants do and hearing what they’re thinking while they do it is what enables the observers to see the site through someone else’s eyes and understand why some things that are obvious to them are confusing or frustrating to users.

  Who should observe?

  As
many people as possible!

  One of the most valuable things about doing usability testing is the effect it can have on the observers. For many people, it’s a transformative experience that dramatically changes the way they think about users: They suddenly “get it” that users aren’t all like them.

  You should try to do whatever you can to encourage everyone—team members, stakeholders, managers, and even executives—to come and watch the test sessions. In fact, if you have any money for testing, I recommend using it to buy the best snacks you can to lure people in. (Chocolate croissants seem to work particularly well.)

  You’ll need an observation room (usually a conference room), a computer with Internet access and screen sharing software, and a large screen monitor or projector and a pair of external speakers so everyone can see and hear what’s happening in the test room.

  During the break after each test session, observers need to write down the three most serious usability problems they noticed during that session so they can share them in the debriefing. You can download a form I created for this purpose from my Web site. They can take as many notes as they want, but it’s important that they make this short list because, as you’ll see, the purpose of the debriefing is to identify the most serious problems so they get fixed first.

  What do you test, and when do you test it?

  As any usability professional will tell you, it’s important to start testing as early as possible and to keep testing through the entire development process.

  In fact, it’s never too early to start. Even before you begin designing your site, for instance, it’s a good idea to do a test of competitive sites. They may be actual competitors, or they may just be sites that have the same style, organization, or features that you plan on using. Bring in three participants and watch them try to do some typical tasks on one or two competitive sites and you’ll learn a lot about what works and doesn’t work without having to design or build anything.

  If you’re redesigning an existing site, you’ll also want to test it before you start, so you’ll know what’s not working (and needs to be changed) and what is working (so you don’t break it).

  Then throughout the project, continue to test everything the team produces, beginning with your first rough sketches and continuing on with wireframes, page comps, prototypes, and finally actual pages.

  How do you choose the tasks to test?

  For each round of testing, you need to come up with tasks: the things the participants will try to do.

  The tasks you test in a given round will depend partly on what you have available to test. If all you have is a rough sketch, for instance, the task may consist of simply asking them to look at it and tell you what they think it is.

  If you have more than a sketch to show them, though, start by making a list of the tasks people need to be able to do with whatever you’re testing. For instance, if you’re testing a prototype of a login process, the tasks might be

  Create an account

  Log in using an existing username and password

  Retrieve a forgotten password

  Retrieve a forgotten username

  Change answer to a security question

  Choose enough tasks to fill the available time (about 35 minutes in a one-hour test), keeping in mind that some people will finish them faster than you expect.

  Then word each task carefully, so the participants will understand exactly what you want them to do. Include any information that they’ll need but won’t have, like login information if you’re having them use a demo account. For example:

  You have an existing account with the username delphi21 and the password correcthorsebatterystaple. You’ve always used the same answers to security questions on every site, and you just read that this is a bad idea. Change your answer for this account.

  You can often get more revealing results if you allow the participants to choose some of the details of the task. It’s much better, for instance, to say “Find a book you want to buy, or a book you bought recently” than “Find a cookbook for under $14.” It increases their emotional investment and allows them to use more of their personal knowledge of the content.

  What happens during the test?

  You can download the script that I use for testing Web sites (or the slightly different version for testing apps) at rocketsurgerymadeeasy.com. I recommend that you read your “lines” exactly as written, since the wording has been carefully chosen.

  A typical one-hour test would be broken down something like this:

  Welcome (4 minutes). You begin by explaining how the test will work so the participant knows what to expect.

  The questions (2 minutes). Next you ask the participant a few questions about themselves. This helps put them at ease and gives you an idea of how computer-savvy and Web-savvy they are.

  The Home page tour (3 minutes). Then you open the Home page of the site you’re testing and ask the participant to look around and tell you what they make of it. This will give you an idea of how easy it is to understand your Home page and how much the participant already knows your domain.

  The tasks (35 minutes). This is the heart of the test: watching the participant try to perform a series of tasks (or in some cases, just one long task). Again, your job is to make sure the participant stays focused on the tasks and keeps thinking aloud.

  If the participant stops saying what they’re thinking, prompt them by saying—wait for it—“What are you thinking?” (For variety, you can also say things like “What are you looking at?” and “What are you doing now?”)

  During this part of the test, it’s crucial that you let them work on their own and don’t do or say anything to influence them. Don’t ask them leading questions, and don’t give them any clues or assistance unless they’re hopelessly stuck or extremely frustrated. If they ask for help, just say something like “What would you do if I wasn’t here?”

  Probing (5 minutes). After the tasks, you can ask the participant questions about anything that happened during the test and any questions that the people in the observation room would like you to ask.

  Wrapping up (5 minutes). Finally, you thank them for their help, pay them, and show them to the door.

  A sample test session

  Here’s an annotated excerpt from a typical—but imaginary—test session. The participant’s name is Janice, and she’s about 25 years old.

  * * *

  Introduction

  Hi, Janice. My name is Steve Krug, and I’m going to be walking you through this session. Before we begin, I have some information for you, and I’m going to read it to make sure I cover everything.

  You probably already have a good idea of why we’ve asked you to come here today, but let me go over it again briefly. We’re testing a Web site that we’re working on so we can see what it’s like for people to use it. The session should take about an hour.

  I want to make it clear right away that we’re testing the site, not you. You can’t do anything wrong here. In fact, this is probably the one place today where you don’t have to worry about making mistakes.

  We want to hear exactly what you think, so please don’t worry that you’re going to hurt our feelings. We want to improve it, so we need to know honestly what you think.

  As we go along, I’m going to ask you to think out loud, to tell me what’s going through your mind. This will help us.

  * * *

  I’m reading from the script that I use when I do usability tests.

  You can download this script at rocketsurgerymadeeasy.com.

  * * *

  If you have questions, just ask. I may not be able to answer them right away, since we’re interested in how people do when they don’t have someone sitting next to them to help, but I will try to answer any questions you still have when we’re done.

  And if you need to take a break at any point, just let me know.

  * * *

  It’s important to mention this, because it will seem rude not to an
swer their questions as you go along. You have to make it clear before you start that (a) it’s nothing personal and (b) you’ll try to answer them at the end if they still want to know.

  * * *

  You may have noticed the microphone. With your permission, we’re going to record what happens on the screen and what you say. The recording will be used only to help us figure out how to improve the site, and it won’t be seen by anyone except the people working on the project. It also helps me, because I don’t have to take as many notes.

  Also, there are a few people from the Web design team observing the session in another room. (They can’t see us, just the screen.)

  * * *

  At this point, most people will say something like, “I’m not going to end up on america’s Funniest Home Videos, am I?”

  * * *

  If you would, I’m going to ask you to sign a simple permission form for us. It just says that we have your permission to record you, but that it will only be seen by the people working on the project.

  Do you have any questions before we begin?

  No. I don’t think so.

  * * *

  Give them the recording permission form to sign.

  You’ll find a sample form and some other useful forms and checklists at rocketsurgerymadeeasy.com.