Having fun analysing conversions?

I was asked at a recent network event about what I would recommend (and what do we do at I-COM) to test how a website is converting.

I know from years of experience that the individuals and companies that are the most successful with their online marketing do one thing more than most - they are constantly testing and tweaking their sites, always on the lookout for changes they can make to improve their results even by a tiny fraction!

Testing is easy to do but I’m not surprised that so few people actually do it! Why? Because it takes time. What are we chasing most of the time? Yes, the clock! So, these great ideas, testing different headings, graphics, call to action banners, content never gets implemented... true?

So what is going to help? Well we have these fundamental rules at I-COM as far as testing is concerned; it's not rocket science but having a process helps.

1. Test one thing at a time!

The key to successful testing is to quickly identify each new element that makes a difference to your conversion rates - if you changed your headline, your content and your call to action all at once, how would you know which of these changes was making a difference (either good or bad)? It sounds like common sense but how many of us break this rule if we are honest?

2. Be methodical!

Slowly and methodically go through your website, testing one element at time, monitoring the results carefully, and then either keeping or removing each change, before you move on to the next test.

In this way, over time, even tiny improvements will add up to a large increase in conversions and sales!

3. Testing should be fun!

It should be fun to test on a regular and ongoing basis, some over weeks, others over months. Look forward to reviewing the results and making the next set of changes as this will keep you motivated and surprise, surprise you will find the time.

4. Benchchmark to measure your test!

Look at your current website stats from Analytics or whatever performance tool you are using. Decide on the pages you are going to measure and change. Don’t go mad - the more you do the more you will have to measure, and what did we say about time or lack of it?

So then, make one change, whether that is to graphics, navigation, headlines or content on each of the pages that you have decided to measure.

For example, if a call to action grahic on the page was generating 1% click through conversion rate and your alternative call to action banner generated a 4% conversion rate, you would leave it and that figure becomes the new benchmark. The next time you test the call to action banner the aim is to try and beat that new one.

The game begins and it’s rewarding and fun and creates a positive behaviour of consistent improvement.

5. Timing!

Ideally you should run tests within the same time frame to avoid skewing your results.

For example, you shouldn't test one headline in September and another in December when seasonal factors could make a difference to your sales. Or, if you're testing an email, you should send the two test emails to the same number of people to get a more accurate idea of which performs better.

The other important thing to remember is to always keep a record of your test results!

As far as I'm concerned the key numbers to keep an eye on include your bounce rate (number of people who leave the site again without visiting beyond the first page they see), number of visitors who reach your order form, your sales conversion, opt-in and enquiry rates.

The truth is, your job of testing never ends... but make it fun!