To achieve people-based marketing, we have to understand customers on a one-to-one basis (or as close as possible). Naturally, audience modelling has seen progression in its sophistication from opinion-based to rule-based and now increasingly to statistically modelled audiences, as we move to take advantage of machine learning technology. Lifetime value analysis such as RFM can provide quick wins in driving lead value and stimulating remarketing activity.
RFM stands for Recency, Frequency and Monetary and the analysis is used to estimate a customer’s value to a business, all from said business’ first party data. We model these three variables to score a user’s value by addressing the following questions:
- Recency - When was the last time the user was last on site?
- Frequency - How regularly do we see this user on site?
- Monetary - How much money does this user typically spend with us?
With these three data points we rank users from high to low creating a combination of 27 different audience segments that we can apply a varying degree of targeting strategy behind.
Naturally, the monetary element lends itself to an eCommerce focus, however this model can be applied to non-eCommerce businesses by substituting revenue with a goal conversion or lead generation value instead.
Let’s Get a Bit Technical
Sophisticated audience segments call for sophisticated technology stacks. If Google Analytics 360 is your analytics platform of choice, you’ll need to be collecting user IDs as a custom dimension on your site to be able to match returning visitors.
Already have this up and running? Well one of the benefits of GA360 is that the integration with Google BigQuery is seamless and you can manage and sort your data into the appropriate high, medium and low splits in the cloud platform.
Personalisation will stem from being able to leverage your data using this model, to qualify the value of customers who purchase (or convert) on your website. The idea behind this is that if you know that ‘X Users’ spend a lot of time on site, but haven’t visited in a while and are previously known to spend a lot, then we should look to provide a tailored response to these users, which could be in the form of an updated message on the latest offerings for your brand.
‘Y Users’ spend a short amount of time on site, visit a lot and haven’t spent very much in the past. These users should naturally be shown a different message to ‘X users’ as they do not have the same response signals and naturally represent different points of value for a client.
The strategy that a client ultimately takes will depend on what KPI targets they wish to meet, but across all strategies a personalised response should be at the centre of any people-based strategy. If you’d like to know more about this model, the implementation or the strategy behind it, contact the Media Science team at Merkle|Periscopix. For more information on people-based marketing, click here.