Monthly Archives October 2014

Testing Your Website: Are You Doing It?

Testing in an educational setting

Testing is something everyone with a website should be doing, but almost nobody does.   It’s the crucial activity to take the guesswork out of your site, whether you’re a blogger, a small business, or a nonprofit organization.  It impacts your written copy, your design…everything. I believe this so strongly, my professional recommendation is that you slap together a non-embarrassing website for cheap (more on this in another post), and then start testing different options.  Don’t hire a photographer or a web designer, or a developer until you’ve done your testing to discover what works for your audience.  Then you can tell them what you actually need.

Why should I be testing on my site?

Here’s why:  You’re just not going to guess correctly on a regular enough basis to be useful.  Professional designers and developers will probably be able to implement things that are more likely to get it right, but still not often enough.  People basically suck at guessing what’s going to work.  I’m talking to you.  And I’m talking about myself.

An Example: Testing vs. Guessing

To prove my point, here’s a little quiz: Let’s say you have a registration form you want people to fill out.  You wonder if adding some text about privacy will improve registrations.  Do you:

  1. Add “100% Privacy — we will never spam you!”
  2. Add “We guarantee 100% privacy. Your information will not be shared.”
  3. Leave it alone — nobody’s complained about it, so don’t mess with what isn’t broken.

Give yourself a moment to choose what you would do.  This process is how most companies go about making decisions about their website.  They come up with some ideas and pick one.  In this case, these options are actually from a real series of tests that were written up as a case study on Unbounce.com, meaning you can find out the “right” answer without doing the test.

But first you have to guess.  There’s a 33% chance of getting it right randomly (even less so if you were to brainstorm more options).

Now for the data

For starters, the third option is what the case study already had in place, so it was the control.  No change in what you’ve showing means no change in your results.  If you chose the first option, you have done yourself a disservice.  That change reduced registrations by almost 19%.  But the second option on the list (the one that doesn’t mention spam) actually increased registrations by nearly 20% compared to the control.

So should you add a comment about privacy to your form?  Maybe…and maybe not. You won’t know for sure unless you do these statistical tests and see what impact they have.  I’ve done enough tests (like the coupon testing I talked about in a previous post) and read enough case studies like the one this example is from to know that sometimes big changes make no difference at all, and tiny changes can have a huge impact.  I saw one example where simply changing the color of the background, or a single word on a button, or the number of bullets in a list either boosted the goals of the page (or made them plummet!)

Test first.  Get the real answers.

THEN go out and tell your developers and designers what they need to do.  Because, with data in hand, you’ll know.  You’ll have eliminated the guesswork.

Posted by Michael J. Coffey  |  0 Comment  |  in Analysis & Testing
  • Stay Connected