Using real user monitoring to put a value on lost conversions

How does your website’s speed and reliability affect conversions? Our real user monitoring (RUM) service captures and analyses conversion data. And it’s already proving valuable to customers using the service.

One customer thought their website had some problems during a big promotion. It was based on anecdotal evidence, and they weren’t initially sure about either the scale or the impact of the issue.

Nevertheless, they were worried: they were running a large promotion and had invested in multi-channel marketing to support it.

So we decided to investigate.

First we looked at performance. Was the anecdotal evidence backed up by hard data? The answer, it turns out, was yes, and we saw a number of obvious slowdowns over the period (represented by the peaks on the graph below).

perftrends

Next we looked at the Impression Trends report to see if there was any obvious fall in impressions over the same period. And it looks as though there were (we’ve circled the relevant parts of the charts in red).

rum1

Day 1

rum2

Day 2

Having seen that there was a significant drop off in visitors over some periods, we wanted to understand whether and if so how it had impacted business metrics, specifically conversion (as defined by the customer).

The chart below shows the number of conversions in 15 minute buckets, with the number of conversions during this promotion much higher than on a typical day…

rum3

Conversions were much higher while the promotion was running.

However, if we focus on the Saturday and Sunday, we can see that there are some serious anomalies (highlighted in yellow in the graph below). As you might expect, drops in the number of conversions lined up with the exact same periods we noticed the drop in impressions!

rum4

A drop in overall impressions was reflected in a drop in conversions

Based on this data, we were able to ‘fill in the gaps’ and estimate that the performance issue lost the customer about 600 conversions during this period. The customer was then able to use its average order size to assign a financial value to those lost conversions. This is a huge piece of insight for them!

So with the help of RUM, we were able to identify a performance issue, see its effect on visitors and quantify the impact on revenue.

Finally, it’s worth noting that the problem looks to have been caused by the weight of traffic and should therefore have been avoidable, not least through load testing.

If you’d like to get similar insights from our, you can learn more and sign up for a free trial here

Leave a Reply

Your email address will not be published. Required fields are marked *