GA4’s ‘Engaged Sessions’ is a terrible KPI and potentially a very misleading way to measure engagement.
a) it’s a binary flag – this session was engaged or it was not. No indication of whether that visitor looked at 2 pages or twenty nor how long they looked at the them.
b) looking at lots of pages or spending lots of time on them may be an awful indication of engagement – it depends on your business. Maybe the visitor is ‘engaged’ because they’re lost and are trying to find the content that is buried in your site.
Not all content on your site is equal – a customer reading about how to unsubscribe from your service or cancel their order is not a good sign of engagement. Someone browsing dozens of pages on an e-commerce site trying to find the product they want is not a great sign.
Is there a better way to measure the engagement of the traffic you are driving to your site?
Yes.
Get hold of the raw data (activate a data transfer into BigQuery if you’ve not yet done this) and start to categorise your pages according to the value they deliver to your business and play in your sales funnel. Simplistically, a product detail page or demo page is worth more than a homepage view, someone spending lots of time on the demo page vs sometime spending very little time on it is worth more to you – then you can start to build a metric which quantifies the time on a page, the value of the page and whether or not a lot of time on that page is a good or bad thing.
This approach allows you to create a score at a visit / session level (more on Visits v Sessions in a future post…) and then you can roll these scores up by the source / campaign / channel level and you have the basis for a new KPI – call it something like ‘Modelled Visit Value’ and you can then look at this on a per session basis, bring your Ad spend into reports alongside it to do a ROAS-type calculation.
Suddenly you have a vastly better way to measure the ‘engagement’ of the traffic you’re buying and directing to your site, and more robust basis on which to optimise your tactics.
It requires a bit of engineering and data wrangling to get the raw data, and you want to ensure the parameters of your model are easily updated so you can deploy it, analyse the results, optimise your spends and re-assess.
As always being able to iterate quickly and painlessly is key to improving performance : there are no silver bullets!
Get email updates
More insights from Bright Analytics