Google Analytics Custom Metrics Caveats
Suppose every week I want to calculate the lifetime value of all my customers, and send that data into Google Analytics (Universal) via the Measurement Protocol.
- I set up LTV as a User level custom dimension.
- I set up LTV as a Currency custom metric.
- I sent in some test data.
- I created a custom report, as shown below
User 10 has a LTV of £100, and User 3 £599. You can see this in the custom dimension.
The trouble is that because I have set the LTV multiple times for each User ID (as it updates each week), the LTV metric is incorrect. Google Analytics has summed each value it has seen for that customer, which is the default behaviour of metrics, but not what I want.
I can’t use this metric for anything useful.
Let’s look at the custom dimension value. This is ok (because I set it a User scope Google Analytics will automatically override all previous values ever seen for this user). But I can’t used it to create a segment that says ‘Show me everyone who’s LTV is > 100’ because the dimension is a string not a float.
It is also difficult to create a report that shows average LTV by traffic source, which is ultimately what I want to do.
(NOTE: Don’t get hung up on LTV. This could be any metric which changes over time and you need the last value, e.g. NPS or customer satisfaction score. The same issues apply)
Assessing the options
The option to get around this problem are to export all GA data and creating the report outside of it, which bring up two problems:
- Log level export from Google Analytics requires Premium (not the free version) and BigQuery. Google haven’t sorted out the contracts either so this option isn’t available in the UK yet.
Why use Google Analytics in the first place? If you push data into Google Analytics just to take it out a while later, what’s the point in putting it in their in the first place. Sure you have some pretty reports, but if you can’t analyse the metrics that really matter to your business, why bother with it?
A platform such as (KISSmetrics)[http://www.kissmetrics.com) might be more suitable. Or, as I am beginning to advice my clients to do, is to create a more custom built data management platform that addresses their specific needs.
Need expert Google Analytics help? Why not consider hiring a Google Analytics consultant?