Getting Started With Segmentation


It can be daunting to face the challenge of making the most of customers in transactional databases. This challenge is made worse by that shiny new marketing automation license sitting up in the cloud, peering down like a Grecian god, targeted campaign thunderbolts in hand.
Service messages can go to the whole base of course, but blasting the same offer to everyone, because you can, just feels wrong. And it is.
We are still believers in the original CX marketing mantra, IDIC:1


Identify customers individually

Differentiate them based on

  1. Their value to you
  2. What they need from you

Interact with them and learn from each exchange. Remember.

Customise your treatment of the customers based on what you have learned about them, and hopefully them about you.

Sounds easy, often it is not.


Ellipsis Guide To Getting Started

Assuming you know the identity of your customers through contracts or a loyalty program that allows you to track interactions and transactions over time, 1to1 relationships may be your eventual goal, where your level of insight is so granular that you can truly personalise interactions with each and every customer. For many businesses, getting to this level of detail is still uneconomic, so segmentation is a way of moving along the ‘segment of one’ continuum as rapidly as costs allow.


1.The Importance of Segmentation

Segmentation is certainly not a new concept, but it’s odd how many companies invest in complex data systems yet still struggle with understanding who their customers are, what value they bring, and what future value they could potentially generate. Often, the initial segmentation will be based around answering the ‘who’ question: who are my customers, where do they live, what’s their income, what’s their family size, which media channels do they use, and so forth.
The next step is commonly based around answering the ‘what’ question: what are my customers doing. For example, the classic RFV (or RFM) approach considers recency of the last visit, frequency of visit, and monetary value of those visits. And value-driven segmentations can go much further, even to the point where computers can be programmed to observe customer behaviour, independently derive their own micro-segmentations through ‘machine learning’, and execute marketing activity based upon this learning, with minimal human interference.
There are also a number of empirically based ‘rules of thumb’ that we see in retail marketing, that are helpful to kick-start our planning. One is that it is more productive to market to increase frequency (get customers to shop more often) than to try to get customers to spend more each visit. The second is that timing is more important to marketing success than the content or richness of the offer.
Understand the rhythm of your customers’ purchases, identify that group within each segment who are most regular in their inter-purchase timings because these customers can be enrolled in a communications plan that matches their individual replenishment cycle. Include new customers who are behaving the same way as high value customers (what they buy, when etc.) as ‘high potentials’.


Combining the ‘who’ and the ‘what’
Ultimately, good segmentations combine the ‘who’ and the ‘what’, the art and the science, to build a rounded view of the customer. Why do we need both? Because fundamentally, segmentations need to do 2 things:


  1. To make the customer understandable to management and staff – this is where the ‘who’ helps, talking in the everyday language of homes, jobs, families and lifestyles.
  2. To be useful in answering business questions that help the business grow – this is where the ‘what’ helps, identifying customers with high churn, looking for cross-sell opportunities, or coming up with ideas for new product development. 


So combining the ‘who’ and the ‘what’ helps to build these rounded segments, often called customer ‘personas’.
And what does useful look like? Coalition programs, for example, often deploy scoring mechanisms for how engaged members are with the overall program, as well as customised segmentations to understand how members save up and spend their points, or how they interact with various program communication channels. These are segmentations which answer the ‘who’ – helping staff understand members – as well as the ‘what’ – shaping marketing activity and business decisions.
It’s important to recognise that even the best segmentations are still by their nature compromises – aggregations of customers into practical chunks to help our understanding and shape our decisions.


2.Develop Personas

For each segment (we generally start with 3-4 until we know more) construct a persona that describes what the typical member of the segment looks like. Personas are a heuristic device that helps team discussions, so don’t aim for precision, look to differentiate.
It is important to remember that perfect is the enemy of good. It’s better that your ‘personas’ are clear and comprehensive, than accurate but awkward; better they are useful and can be acted on, than pristine but impractical outputs of data analysis.
Where to start
Ask for each persona; ‘What do we want from these customers?” This can be as simple as;
  • Retain HV customers (they are already buying all they can so recognise them)
  • Increase shopping frequency from MV customers (get them to consolidate their shopping with us)
  • Encourage a second purchase from the LV customers
Then decide, for each persona; “What do these customers want from us?”. The toughest part of the process is determining what does each audience want from us that is not ‘cheap but good’. It may be a different service, product configuration, timely access, personalised communications & offers, auto replenishment… contemplate your customers and be creative. But do start where the largest economic opportunity is in the customer base.
Make your communication strategy explicit
Make your strategy for each persona clear. It helps to think about communications / interactions at three levels;
  • Immediate, in the moment e.g. recognition at check out, priority call numbers
  • Triggered communications in response to a customer event e.g. they are late coming back so prompt them with an offer
  • Relationship communications intended to convey to the customer that you are not only interested in selling them stuff but that you are also paying attention to how they interact with you.
Understand Interactions
Finally, pause and look at how the segments influence each other, if at all. Social channels can be potent forces in your relationship with specific (or all) customers. Are there advocates who can be amplified, are there disgruntled influencers who should be addressed?


3.Invest in Technology and Data

It’s important that right from the start, you align each segment with a set of success metrics that are specific to your strategies and make sure you have the data sources and systems to deliver these measures regularly and accurately.
Technology will ultimately provide this level of insight and reporting, getting us back to the ‘corner store’ paradigm where the shopkeeper knew every customer personally, and what they needed. The idea is that through technology, a memory builds up over time, allowing the business to remove friction from customer interactions because needs are understood and can be anticipated. This increases switching costs for customers who, when defecting to the competition, would have to become strangers again, and lose their 'comfortable' relationship with their supplier.
We see some examples of this already in the market, as businesses increasingly investing in data science to fully exploit the potential for deeper customer relationships2. There are fitness centres which track interactions between customers and instructors, allowing the instructors to emulate the behaviour of long term relationships by referencing previous activity. There are casino operators who provide their front-line staff with live customer data, so they can provide relevant real-time support based on actual gaming behaviour. And of course, there are the grocers who use billions of rows of transactional data to develop personalised offers by curating the most relevant deals for each individual shopper.
As spectacular as advances in technology may seem, we should not lose sight of the fundamental purpose of these developments: to help us understand our customers better, to provide higher quality interactions, and make better business decisions.


Final Thoughts

Remember, before jumping into the deep end of expensive data systems and analytics, to go back to fundamentals, and to define the core business questions, challenges or gaps in customer understanding which you need your insights to solve.
Hopefully this is enough to get started, with some reassurance that the path to more effective CX is measured and methodical. Those pesky customers will not hibernate however, they will change their views and brand relationships, so revisit your strategies regularly!


We are Ellipsis, the Customer Loyalty Experts. We help businesses thrive through solving complex customer problems. Please get in touch, we’d love to talk.



Peppers and Rogers Group

Mckinsey, The dawn of marketing’s new golden age, Feb 2015

Download the White Paper

Fill in your details to download this white paper.

Keep me updated with the latest news
Sign Up to receive our latest news and white papers