Don't Know Why Conversions Are Low? Just Ask Your Users!

This is part five of a multi-part series on The Infinite Conversion Loop. If you have not read the previous articles, you should check out Introducing the Infinite Conversion Loop, 10 Things to Check in Your Google Analytics, Who are these people visiting my website?, and How to Legally Spy on Your Visitors.

Last time we talked about Passive User Testing, that is, examining user behavior without their knowledge or involvement. Passive user testing is great for finding large problems, but it’s hard to get the “why” from this method of testing. For example, you may notice everyone is abandoning at the first step of the shopping cart, but if there is not a JavaScript error or something obvious, it may be hard to fix this problem.

Active User Testing is the best way to get more answers. It’s more expensive then passive user testing, which is why I recommend it only after you’ve fixed the real big problems. You really don’t want to waste money on people telling you the site doesn’t work at all in Internet Explorer (which can easily be identified using automated means). Active user testing involves (usually) paying people to use your website and verbalize what they are doing.

If you’ve never done this type of testing before, the best primer for getting started with user testing is Steve Krug’s excellent “Don’t Make Me Think“. There are a couple of ways to accomplish user testing, some companies have dedicated labs with two-way mirrors, complicated video setups, etc… but you don’t need to go to this level to get some valuable insight.

The first step to user testing is coming up with a script. This can be the hardest part, but it’s important you come up with a realistic scenario that people can identify with. Don’t say things like “click the upper nav bar and go to the shopping cart”, instead tell the person “You are a parent shopping for Christmas. You’ve just landed on Amazon.com and need to purchase a remote controlled car. Please narrate your thoughts while you attempt to accomplish this task”. Don’t provide any assistance or answer questions unless the person is so stuck that the test cannot continue.

By far the easiest and fastest way to get started user testing is using UserTesting.com. They already have the setup for recording people’s screens and they have people that are ready to user test literally at a moment’s notice. You typically request users from their pool of users based on specific demographics such as “novice computer user” or “female professional”. It should be noted these demographics are self-selecting, and I always have a little doubt that truly amateur computer users will have signed up as online user testers, but for the most part I’ve had very good experience with this service. It can get pricey though (10 users costs ~$400). In the end you’ll get a video of the user trying to accomplish your steps with their voice narrating everything.

A more economical way of doing this is to literally pull people in off the streets and pay them $10 or offer free pizza. This is one way to get completely average people, however the negative is they will be using your equipment, so it’s more difficult to rule out browser/OS specific issues. You’ll also need to setup a method of recording them and their behavior (a web cam and screen recording software is really all you need).

Any way you’ll do user testing I can almost guarantee you’ll get some insights that never occurred to you before. People will say things like:

“There was no BBB logo on this site, so I didn’t trust it with my credit card”
“It asked for a fax number, but I don’t have a fax so I cancelled”
“I can’t find a way to search for X”
“It wouldn’t show me the shipping amount before I gave it my card number, and I was afraid you’d charge me”

You’ll notice many of the examples I’ve given involved payment information. Giving credit card info to your website it’s the biggest sign of trust users have, and this is where I almost always hear interesting insight from customers as to why they didn’t trust giving payment information.

We’re almost complete the Infinite Conversion Loop. Next time I’ll talk about one of my favorite topics, A/B Testing. Stay tuned!