The Blog of Scott Chapman
Copywriter & Sales Conversion Expert

4 points you must remember during conversion rate split-testing

  • Add to favorites
  • Google Plus
  • Twitter
  • Facebook
  • LinkedIn
  • Digg
  • Diigo
  • Fark
  • NewsVine
  • StumbleUpon
  • Ping.fm
  • Technorati
  • Slashdot
  • Reddit
  • Pinterest
  • RSS
  • Netvouz

A while back, after a round of successful split-tests, my client asked me this question:

“How do you do it?”

He already knew about the software and platforms I use, but he wanted to know exactly what I do to ensure that split-tests build towards higher conversion rates and more sales.

After all, the very nature of A/B testing is a pretty reliable way to increase website income, but it can go wrong, or at the very least, progress can be slow.

Here are my 4 main points to remember when conducting any website split-test:

1). Don’t act too soon

Some split-tests can go through what I like to call a “honeymoon period”. Often, within the first days of a test, conversion rates can shoot up, giving the impression that the test is well on its way to giving you a winning variation.

However, in some cases, after more days have passed, the conversion rate gradually falls back to baseline, and on the odd occasion, it can even fall below baseline and give you a worse conversion rate than you started with.

This can be surprisingly common. A winning variation with 200 visitors can become a losing variation after 1000. It works both ways, too. A variation which seems to be losing initially can return an increased conversion rate after more time.

This is why it’s vital to give tests enough time to build statistically significant results. If you use software which highlights statistical confidence, then it should be no less than 95% before you declare a test complete.

Another question frequently asked is how much traffic do you need for A/B split tests? There’s no universal answer to this, but the more the better. At the very least, most tests would benefit from a minimum of 1,000 total visitors, but preferably more.

2). Heavy-hitters first, marginal improvements later

There are always elements of a landing page which will have the largest impact on conversion rates. For example, changes to the main headline and opening paragraphs will have much more of an influence compared to button text and colours.

This is why it’s important to start with split-testing the heavy-hitters, as you’ll get larger initial improvements and unlock the potential of extra sales sooner rather than later.

Once conversion rate improvements start to slow, it’s time to put more emphasis into the marginal improvements, such as split-tests involving small text tweaks, colours and layout adjustments.

These should not be underestimated, as they can all accumulate into a sizeable conversion rate increase. For example, ten 0.1% conversion rate improvements give you a 1% increase overall, and if that results in your website’s conversion rate going from 4% to 5%, that’s a 25% increase in sales with no additional marketing spend.

3). Focus on the metrics which really matter

Bounce rates. Button clicks. Time on site. There are plenty of statistics to look at, but they should never be allowed to cloud your judgement. The statistics which matter most are those which are directly responsible for how much money you make, namely metrics like conversion rates and cost-per-conversion.

However, even these aren’t always the be all and end all. It’s important to also look at your overall sales and profit levels. For example, a split-tested variation could increase the conversion rate and purchases, but reduce average order value, so you ultimately end up making less profit overall.

4). Apply proven conversion rate optimisation knowledge

Winning variations will simply happen more frequently if you apply the proven principles of good conversion rates.

The very nature of A/B testing means that you’ll be guided down the correct path no matter what you test (i.e. bad ideas will always lose), but it helps to do your research and test what is known work well in the world of CRO. It makes the chances of finding winning variations more likely so you can make faster progress in the right direction.

Keep these 4 tips in mind when conducting split-tests on your website, and you should see faster, larger and more consistent sales improvements.


  • Add to favorites
  • Google Plus
  • Twitter
  • Facebook
  • LinkedIn
  • Digg
  • Diigo
  • Fark
  • NewsVine
  • StumbleUpon
  • Ping.fm
  • Technorati
  • Slashdot
  • Reddit
  • Pinterest
  • RSS
  • Netvouz

Share and Enjoy

Posted in Copywriting, Marketing | Leave a comment

Comments are closed.

 

Comments are closed.