20% of our sales are from AdWords, so there is a 20% chance our PPC visitors make a purchase. WRONG.
This is a classic mistake made by a wrong understanding of statistics and often leads to wrong marketing investment decisions. Within the field of Internet marketing it generally results in Paid advertising or email marketing being attributed too much credit and non- paid search work not being valued to its full worth.
The mistake arises because an analysis is made using the sales data only and not the website data. Increasingly CRM systems are importing some marketing channel data, and making it easy for these mistakes in marketing analysis to happen.
On a typical eCommerce website, if 5% of your visitors convert to sales, and just 10% of your visitors are from AdWords, then if 20% of sales are from AdWords there is a 10% chance a PPC visitor will make a purchase.
But if half (50%) of your visitors are from AdWords, the chance that one will make a purchase is just 2%. If all other sales are from natural SEO, the chance one of the SEO visitors makes a purchase is higher at 8%.
These figures are calculated using Bayesian statistics that deals with the chance of something happening given other information. For the examples above the way of working out the chance that an AdWords visitor will make a purchase is as follows:
Chance that an AdWords visitor makes a purchase = (percentage of sales from AdWords * website conversion rate) / percentage of website visitors from AdWords.
These are all data easily obtainable from programs such as Google Analytics.
- percentage of sales from AdWords = sales from AdWords / total sales
- website conversion rate = no of sales / no of website visitors
- percentage of website visitors from AdWords = no AdWords visitors / total no of visitors
This type of modelling can be extended to look at the value of the sales and ROI predicted.
For those interested in mathematical modelling, the basic Bayes law usually written as:
ExtraDigital use a range of mathematical modelling and statistical techniques to analyse and interpret digital marketing data for our clients.