Blog posts tagged website optimisation

Why you should take testing seriously online

At Econsultancy's Online Marketing Masterclass event in London on Wed 18th November, I listened to an excellent presentation by Craig Sullivan on website optimisation and testing strategies. It may sound like a dry subject but if a single button design change can lead to an increase in revenue of $300m in 12 months, surely dry suddenly becomes crisp and clear?

Firstly, read this article on the $300m button - it gives perspective and is the best way to make you focus on the value of testing.

Why should you take testing seriously?

• Only your customers know what is best for them
• Unless you have a zero bounce rate and 100% engagement your web page can be improved
• Second guessing what customers need can lead to wasted investment
• Web analytics will tell you what is happening
• Customer surveys will tell you why
• Testing will help you optimise the page to improve conversion

How you can run simple testing yourself without having to spend £000s

My take on testing is that you need to prove the business case first - to do this, run simple A/B tests on your worst performing pages using a free tool like Google Website Optimizer. Test page variations, deploy the best performing option. Keep refining your tests until you have improved conversion and proven the investment model.

Once you've proven the business case it then pays to invest in an analytics and testing specialist who can run sophisticated MVT programs on your behalf. If you can manage this in-house and have the time/knowledge, excellent. Don't be afraid of using an expert, they really can deliver ROI. Just make sure you vet them carefully first and ask the right questions. A badly planned and executed testing plan will cost you money and deliver limited returns.

Before you start set your expectations - not every test works. Sometimes you will produce alternative page designs that decrease conversion. Don't panic. A negative test still provides learning and helps you evolve your testing hypotheses.

You can read more on my Econsultancy blog.