As part of our Insights Blog series, Andrew Dunmall, Senior Data Analyst at Infectious Media explains a clients journey to improve their engagement measurement metrics.
Results from our recent survey of over 200 digital marketers show that, despite the ad fraud and industry-recognition of this way of measurement being corrupt, 57% of marketers still use total number of clicks as a key metric to measure programmatic performance.
Not only does post-click attribution massively undervalue the influence of other activity within the funnel, but it also overvalues lower funnel tactics, such as paid search, affiliate marketing, and retargeting.
With this in mind, more and more advertisers have been breaking away from this outdated model, and have been looking towards changing their whole approach to measurement, with outstanding results.
A client approached us acknowledging that their measurement for their online campaigns was unfit. Due to how their product was sold the vast majority of their sales occurred on third-party sites so they couldn’t view the overall impact their advertising was having. The core issue here was that they didn’t have enough data to be able to measure and properly align their campaigns.
The challenge was to provide a metric that would impact sales, whilst not explicitly tracking sales.
How did Infectious Media help improve measurement?
We began by reviewing the data that we had and could use. In this case we used the client’s 1st party data tracked through their website which includes both product information and an online shop. Next we looked to understand how varying data points interacted with sales, and how these may be indicative of the behaviour of a user who will soon convert.
For this client’s product there was a small research period in the path to conversion, which is something we used to our advantage. Taking the data we have on users who are researching products, we can start to compare the different data points we can get from this research phase to see which shows the most intent to buy. For example, does visiting more pages mean you’re more likely to convert, or is it the device you browse on or the amount of time to take to read the content?
Some of these points intuitively fit into a measurement approach, whereas others are important to look at to make sure you don’t reach false conclusions. For instance if you always research on a mobile device but convert on a desktop then you would need to take cross-device into account to see this correlation to sales.
After analysing various different approaches, we noticed that a combination of data points often lead to a far better correlation to sales than any one page or data point. Off the back of this, we developed our session based metric, which is created by combining the various data points of a user that happen in a small time frame, i.e. in one sitting.
What we found
Previously our client was using a minimum time on site (15 seconds) as a metric, which had a correlation coefficient with sales on site of 0.05. We found that in this instance, the length of time spent per session on site correlated eight times better to a sale, whilst still having the advantage of there being more data to help optimise your ad spend against than site sales.
To further improve how actionable this metric was, we were able to calibrate the optimal session length, for volume of data whilst retaining the correlation to sales, to produce a single conversion event that fit closer to the traditional conversion metrics and therefore was easier to implement by our client.
If you take the time to understand the data you have available to you when building out a measurement solution, especially when working outside the realms of direct sales, you can create a more efficient and tailored solution for your brand that will ultimately lead to a better optimised ad campaign, better use of ad spend and overall, more sales.
To find out more about incremental measurement, read our new whitepaper “Measuring what matters”.