Customer segments are the way to even better ROAS
February 5, 2021
If you have a high volume of customers as well as repeat customers, we might just have a tip for you that will allow you to squeeze even higher ROAS out of your dynamic remarketing.
Overall, segmentation is crucial in digital marketing.
Segmentation classically has two purposes: To show the user the right message at the right time.
Dynamic remarketing makes the time easy to work with (from the moment after they have been on the site and a given number of days ahead), and the message is also given (the item they saw, items in the same category, and items of the same brand).
But there is an extra dimension that makes it possible to raise performance even more, and that is about using segmentation to weigh budget and media pressure.
Bid welcome to: The RFM segmentation model.
It provides not only better budget weighting but also obvious differentiations in communication as well as generally really exciting insights about your customers and your customer base.
Use the RFM segmentation model to segment your dynamic advertising
What is the RFM segmentation model?
RFM stands for Recency, Frequency, and Monetary
Customers are segmented based on when they last bought, how many times they have bought, and how much money they have spent.
For example, if you run a clothing store where customers have a high purchase frequency, then you can divide them into Recency in the following:
1-30 days since last purchase
31-60 days since last purchase
61-120 days since last purchase
At least 121 days since last purchase
Similarly, F and M are also divided into four groups.
Each customer gets their own RFM score, where 111 is the best (have bought recently, bought many times and spent a lot of money), while a 444 profile is a customer who bought a long time ago, only bought once, and spent money.
The segmentation is therefore relatively simple but at the same time extremely insightful.
You basically get 64 segments from this division and can advantageously group them. A very simple grouping is based on this visualization, where the monetary dimension is omitted.
The most important segments
The visualization shows a grouping into just six segments. But you can advantageously divide into more if you have the volume for it.
This provides a good overview and understanding of a number of main segments with a uniform description.
In the division, keep in mind that you do not have permission on everyone, so you may only have half with permission, and if the lists are used for Facebook (and Google Search remarketing), they require at least 1,000 matched permissions.
Once you have divided your customers and uploaded the lists, here are 3 examples of how you can improve performance and think about the customer journey in your initiatives:
Best customers segment: If you have a large enough volume, keep this segment to F = 1, ie. those who have purchased within the last 30 days. They have bought many times and recently. So consider how high frequency you should run with your DPA. They will probably have to return, and they may be back in the shop for reasons other than to buy again.
First time buyers but lost/in hibernation segment: This is important. They have bought from you once, so getting them reactivated has tremendous value. They have been past the site again and shown interest, so this segment you need to give gas to get back. I did not want to settle for DPA but also supplement with some brand messages and create security to get sale No. 2 to them at the checkout.
Do not lose segment: This group consists of customers who have performed well in the past, but it's been a while since their last purchase. They know your store really well, so what does it take to get them reactivated? They are already well on their way as they have been past, so prioritize this target audience, consider accepting a higher ROAS to get them back, and go the extra mile in general to get the sale in house. A sale will move them to "Best customers" and you can initiate ambassadorial actions towards them.
If you add the monetary dimension, you can in a more advanced setup assess who is triggered by discounts, who has a high basket size and other things that you can advantageously include in your messages.
How to make segmentation low-practice.
I am a big proponent of a pragmatic approach, so start with a more manual setup, see the value of the segmented bets (which are not only reserved for DPA, but all your advertising towards existing customers), and then plan to automate it.
It can be fully automated, for example by a setup in Google Cloud and BigQuery, where data is updated daily and audiences are shot to Facebook (and Google and your mail platform), but before you get there you can make a manual extract, assess the effect and then update the lists weekly or less frequently.
The manual process is about making an extract of your customer data, where each customer has one row of data.
Enrich this customer with info such as:
Number of purchases
Is there permission?
Date of last purchase (which you convert to days since last purchase)
Data must of course be enriched with which customers you have permission for, so you can upload mailing lists with permission in the various segments and place these lists above DPA.
Two extra bonus tips
Pro tip #1
In addition to the fact that the actual distribution of customers is extremely interesting, also make the distribution in relation to permission, so that you know the share in each segment, with permission.
You might see quite variable leave rates from group to group.
Pro tip #2
Permission is crucial for you to succeed in the above.
Therefore, immediately initiate a lead ad, targeting only users on your receipt page (i.e. buyers) with your current permission list as a negative segment.
This is a focused way to increase the leave rate on customers.