Cleaning Up Your Data: A Simple Case Study

Written by Jon on 17th August 2016

A couple of weeks ago I wrote a blog post about the different types of attribution models Google used to measure conversions, from which we received a couple of requests and comments asking us about the quality of data coming from Google Analytics and what your ACTUAL conversion rates are.

So, that got me thinking and I rang one of my oldest friends who has an ecommerce website in the UK and asked him if he would like to know what his actual conversion rate was and if I could publish my findings.

He (thankfully) agreed on the condition I bought him a pint the next time we caught up, this I felt was fair!

So using the following date range we looked at a number of data points without making any changes to the their Google Analytics Account.

January 1st 2016 – March 31st 2016

Conversion Rate – 5.54%

Unique Visits – 47, 575

Bounce Rate – 52.53%

Average Time on Site – 00:02:19

Removing the Spammy Referral traffic

The first thing we will do is to exclude all of the spammy referral traffic. You can find out exactly how to do that here. So that we could clean it from historical data, we created a segment in Analytics that excluding all the spambots that we’d spotted in the referral report.

As soon as we do this, we are left with the following results:

Conversion Rate – 5.56%

Unique Visits – 47,347

Bounce Rate – 52.38%

Average Time on Site – 00:02:20

So no-one pays for this traffic however it gets in our way and can dilute our decision making.  Data driven decisions are great …….. as long as your data is accurate!

Narrowing your Target Audience

The website only delivers to 14 countries however 95% of the business is generated within the UK and also where 100% of the marketing spend is allocated.

So if we were looking at making any decisions based on our GA data, it wouldn’t be a bad idea to exclude all non-UK traffic as this constitutes 5% of the business and we can look at this separately later.

UK only Traffic and excluding Spammy Referrals

Once we’ve excluded all other geographical locations except for the UK (their main source of orders), we get:

Conversion Rate – 6.05%

Unique Visits – 41765

Bounce Rate – 51.03%

Average Time on Site – 00:02:25

Looking at the continent

5% is not to be ignored and with no marketing effort could demonstrate an area for growth, so it would be good to look at how the data stacks up

Looking at the EU but excluding all countries that the website does not ship to and also, the UK!

Conversion Rate – 5.93%

Unique Visits – 43294

Bounce Rate – 51.28

Average Time on Site – 00:02:24

So what does this tell us?

Data needs to be looked after and you shouldn’t just assume that the metrics in front of you are correct. Your analytics program (Google Analytics) is clever but not that clever, so you should setup segments that are specific to your business so any decisions you make will be based on data accurate to your business.

For example your conversion rate could completely drop and your bounce rate go through the roof, and the reason?  Your website has been hit by a bot!

To make any decisions based on the changes in this data would be a mistake (unless this decision was to go through your GA account and ensure all irrelevent data is setup correctly!) as it would not be a true reflection of how your website is performing.