In this previous article on customer centricity, we explained how important it is that you understand that your customer base is not ‘one monolithic entity’. That all customers are different and that they need to be grouped into different segments who receive different messages & levels of attention.
The credo reads :’If you know who your best/worst customers are, you can invest your resources more sensibly.’ It makes sense because no-one wants to invest equally in their best and worst customers.
Sooo.. Segmentation is 🔑. Glad we agree on this. So let’s read on!
Of course, there are many ways to segment your customer base. Basic segmentation techniques are usually based on :
- Interests & Affinity (likes ‘soccer’ on Facebook);
- Demographical factors (age, gender, location, etc.);
- Behavioral factors (visited website, watched product page);
- Transactional factors (bought product).
This article will focus on segmentation based on transactional factors. There are several reasons why we value transactional data over all other types of data.
First off, transactional data is often readily available within current systems. Not every company tracks the age / gender / interests / last visit / .. of their consumers. But they do track transactions as it is the backbone of their existence!
Secondly, because they track transactional data, this often means that there is also a large historical database available (transactions are tracked from the beginning of the company).
Lastly, transactional data is easily one of the most important types of data you could track & segment upon. It is the ultimate sign of interest & trust when a customer is prepared to hand over his/her money in return for a specific service / product you offer. Logically, this implies that targeting based on these parameters is also really powerful.
This makes sense right?
That is why – when we start working with customers on segmentation – we prefer kicking off the exercise with transactional data. More specifically, we initiate RFM segmentation.
What is RFM segmentation?
A proven marketing model for segmentation based on the historical purchasing behavior of customers:
Recency: How recent was the last purchase?
Frequency: How often do customers make a purchase?
Monetary : What is the total monetary value of the purchases?
By segmenting your customer base on these axes, several interesting target groups are created on which you can focus your marketing efforts more intelligently. For example the following 4 groups:
High-Value customers: Highest Recency / Highest Frequency
Who? : customers who have bought recently and often.
Tactic: Recognize them and give them love. Keep a finger on the pulse and show them your appreciation, for example with privileges – from an invitation to HQ, to early access to new products or services.
Medium-Value Customers: Medium Recency / Medium Frequency
Who? : customers who have bought not that long ago and moderately often.
Tactic: Promote your medium-value clients to high-value clients, for example, by convincing them to choose a more expensive package. An example of this is LinkedIn Premium.
Low-Value Customers: Low Recency / Low Frequency
Who? : customers who have bought a relatively long time ago and not too often.
Tactic: Customer centricity implies that you don’t want to concentrate on these customers. How can you decrease spending on these customers? After all, they don’t generate a lot of value. Exclude them, for example, in specific paid acquisition campaigns and focus on “cheaper” acquisition by e-mail.
First Time Clients: Recency Very High / Frequency = 1
Who? : customers doing business with you for the first time.
Tactic: You have to subject them to an onboarding trajectory in which you encourage them to make a first repeat purchase. Example: provide your customers with a video that contains an instruction manual.
It is important to note that these are just a few of the audiences you can create via RFM modeling. You can get as creative as you like.
What about more advanced segmentation techniques?
With the rise of machine learning techniques, you might wonder whether RFM is still relevant since the calculations are rather basic. The answer is definitely yes. To be fair, predictive models have critical advantages over RFM. RFM by definition only utilizes 3 predictor variables, whereas predictive models can use hundreds or even thousands. But on the other hand, the fees for building predictive models can run in the tens of thousands, while RFM is essentially a do-it-yourself proposition and, as a result, much cheaper.
With more advanced AI segmentation techniques, you often create a black box as well. You have no idea why this customer is better than the other. This isn’t necessarily bad but a lot of people still struggle with believing random output from a machine. But if you have confidence in other predictive models, do give us a call or fill in a form on the website. We are more than happy to assist in these kind of projects. In an ideal world, you’ll let both models run next to each other.
When you are starting out with segmentation on transactional parameters, RFM is definitely the way to go.