Posted to Ben Finklea's blog on May 10th, 2010

Drupal SEO: How to Find Conversion Problems with On-Site Usability Testing

Usability testing is an easy and objective way to find out what problems users are going to have with your site. Although it's used widely, some call it the secret weapon of web design because it is so often ignored by site designers. The reasons are many, from arrogance to lack of client support to funds. However, in this day where a 1% increase in web site usability can mean a 50% increase in sales, do not leave this tool in the box.

"Debates about the thickness of the drop shadow on the navigation tend to fade in importance as soon as the team sees a prospect struggling to find the Add to Cart button". - Lance Loveday & Sandra Niehaus, Web Design for ROI

Are you having trouble with conversions and not sure what the problem is? Join me after the jump to learn how to find conversion problems with usability testing.

Just a few years ago, usability testing meant renting a $5000 per-day lab with two-way mirrors, and expensive eye-tracking software. With the advent of cheap broadband and web conferencing tools, you can roll your own low-end usability lab for less than $2000. It looks something like the following table:

With these tools, you've got everything you need to do basic usability testing. The high-end labs will offer more, of course, but this solution should provide you with enough insight into the experience visitors are having on your web site. Moreover, since it's there whenever you need it, you'll be far more likely to use it when the time is right.

The Design Process with Usability Testing

The macro process should look something like this:

1. Design

2. Usability test the design, round 1.

3. Adjust design.

4. Usability test the design, round 2.

5. Adjust design.

6. Usability test the design, round 3.

7. Adjust design.

8. Launch site.

Each round of usability testing will reveal issues that you need to improve your design.

The Process for Each User

The biggest question I get are: “Who many tests?” Best practices call for five or six subjects per round of testing, – you could go for eight or ten. However, much more than that and it’s time consuming; less than five and you’re risking a bad sample size.

The following is the process that every user follows:

On-Site Testing

On-site testing is great if you have ready access to your target audience, good facilities, and are using a version of the web site that isn't available to the public. It's also great if you want to eliminate technical issues with system setup and broadband speeds.

If you're testing at your location, the process for each test subject will be as follows:

1. Decide what it is that you want to test about the design. Is it the checkout? FAQ? Finding out how much something costs? Tracking an order? Whatever it is, this should be something along the lines that you wish for regular users to be able to accomplish on your site. The list should include five to eight items and might look something like this:

You're a small business owner and you need to fix your widget flange. You search for the term widget flange on Google and one of the results brought you to this page. Carry out the following steps:

• Find the product that will fix your widget flange.

• Pick the color you want and add the widget flange fixer to the cart.

• Find the return policy. How long do you have before you can return the widget flange if it doesn't work?___________

• Check out the following options:

i. Use credit card number 1234-5678-9101-1112 with expiration 12/12 and CVV 123

ii. Use a different shipping and billing address as provided here: (provide some addresses)

iii. Increase the quantity to 2 items

iv. Select ground shipping. How much would overnight shipping cost? $___________

v. Use coupon code 1234 to receive a 10% discount.

Please Note: Do not put these items in the order that they most logically appear your site if you want to test the back button.

• Now log out the site.

• Go back to the front page and check the status of your order. Find the expected delivery date.

• Send an email to customer service asking them to expedite the shipping.

• You are finished. Please let the tester know.

2. Schedule test subjects that match your demographic. Craigslist is a great place to find folks looking to pick up an extra buck. You can offer them anything from free food or free products to cash—$20 to $50 is typically enough depending on the time commitment and travel distance. In a hurry or on a budget? Recruit your friends and family. It's not as valid as a random sample but it's much better than no testing at all. (My wife, a Ph.D. genetics lecturer, is the best usability tester I've ever seen.)

3. Prior to each test, prep the systems. This is where laptops and WiFi come in handy. You can do the prep on both systems in the testing room and then carry the observation computer into the next room before you begin.

• Create the Skype call & test the cameras

• Create the GoToMeeting Conference and test

• Set up the screen recording software on the observer system and test record some video

• On the subject computer, arrange the windows so that you can clearly see everything

• Open the browser to the site you're testing

• Set the screen saver so that the screen is not visible

• Put the systems in two separate rooms so that you're not inadvertently influencing the user

4. Hit record and do a dry run of the objectives to make sure you've got all the information you need (like a credit card number or user login) to accomplish each task. This will also create a control video to show what should happen if everything goes right.

5. When the subject arrives, thank them for their help and have them fill out a basic demographic questionnaire (gender, race, income, occupation, years experience using the web, or whatever you feel would be helpful).

6. Sit them down in front of the system you want them to use. Ask them not to touch anything until you ask them to begin.

7. Begin recording.

8. Briefly explain the site that you're testing. Ask them to think out loud while they're working. That monologue is very valuable to understand the issues they're having. Explain how you'll be observing everything they're doing but are not available to help.

9. Hand them the list of objectives and leave the room.

10. On the observer computer, be sure everything is recording properly. If it's not, fix it but try not to disturb the test subject.

11. Take good notes, noting the start time and time-stamps of the start of each objective and of interesting things that happen. This saves time when you're reviewing the test later.

12. If the subject gets stuck for awhile on one objective, pop your head into the room and tell them to move on. If objectives are dependent on previous items, tell them which one to do next.

13. When they reach the conclusion of the list, thank them for their participation, give them their payment, and walk them to the door.

15. Reset the systems and repeat with each person.

Patterns will emerge that will convince your team about the changes which need to be made.

Reporting the Results

Preparing reports are time-consuming, though, so keep them to a minimum if you can. Because Drupal is so flexible, you may find that just implementing the changes indicated by the usability-testing is all the reporting you need to do. Show them the testing and say something like 'Here are the changes that our usability tests showed would help our site sell more widgets'. Be prepared to back up your findings with the videos, however.

If you do need to provide more data, create short highlight videos based on each objective. This should be relatively easy if you have good time-stamps. Remember, you're not creating a Hollywood movie here, you just want to get the point across.

Usability testing is not for the faint of heart, though! Be prepared for all manner of wailing and moaning from the web design team when you show them these results. Stay calm! The data are on your side. If 80% (or even 50%) of the users had a problem with an element that the designer is in love with then chances are good that it's serious enough to rethink and change.

Thanks For Reading!

Did you find this post entertaining, useful, or interesting? Please repost, retweet, or redistribute to any of the social sites of your choice, and please subscribe to our RSS feed for daily fodder. For every RSS subscription Volacci gets, a kitten earns its whiskers. You like kittens, don’t you? Do the right thing, then. Subscribe.

We also are very interested in what you have to say in response to this blog post. As always we are very grateful for you, our reader, and greatly value your input. Please start a conversation with a comment below.