How to do split testing for marketing analytics

How to do split testing for marketing analytics

The monitoring and analytics is as important as publishing, because if you do not know your results in detail, how can you know whether time and money spent is worthwhile?  This is a discipline in itself and whilst it can seem off-putting, it isn’t rocket science; it can be learnt.

One method involves split testing your content to see what generates activity by e.g. promoting two offers in different directions, or send segmented audiences to different splash and squeeze pages to see which generates sales.  When you know, continue to test ever after by tweaking details until you have mastered the formula for optimising cost per client acquisition. You want to achieve the least output for the greatest impact.

If split testing is new to you, there is now plenty of advice and checklists  to help you hone a strategy that reaps informative and profitable results.   Beware! Depending on the marketing agency’s specialism, you’ll see differing emphasis on methods employed. For instance, technology producers will put greater emphasis on testing velocity via tools.  Get clear about testing methods and how to understand the statistics, then you have all the skills and knowledge needed to make split testing work for you!

The split testing approach applies to various campaign results and is according to Digital Marketer as follows :

  • Identify goals against any key performance indicators (KPI’s) and aim for a minimum of 100 conversions every time you vary an element of a campaign. Digital Marketer recommend 200-250 for ‘higher risk’ campaigns.
  • Prioritise tests, then gather data, (particularly successful and under-performing pages)
  • Analyse data and seek feedback from users
  • Form assumptions and hypotheses
  • Design and prioritise variants for scalable tests
  • Schedule subset tests to complete within 4-6 weeks and stick to the schedule. Test for full weeks and in alignment with your buyer cycle.
  • Implement or deploy tools, with awareness of potential margin for error (“confidence interval”)
  • Review results and apply new knowledge of ‘lift’ achieved in sales
  • Repeat with new variants and to more subsets for seeing percentage ‘lift’ in conversions. (N.B. A 10% lift is an increase from 1% to 1.1%, not from 1% to 11%)
  • When satisfied move to new content

Many businesses are put off by the maths of analytics, but with time, it will make sense.  For instance, if you know that monitoring the conversion rate of a winning variant in a split test after completion is vital to understand rates of lift, you will begin to recognise buyer patterns and hone into what your clients are looking for and learn more about diminishing returns.  Remember, split testing is only one tool to use while in the process of optimising conversions!


If you’re looking for premium quality copywriting, web content writing and blogs in Stevenage and Welwyn Garden City, send us a message here or call us now on 07507 295566.

No Comments

Post A Comment