Segmentation model or segmenting is a way to divide your customers into sub-groups. There are different ways of segmenting and different reasons to segment. Let’s dive into a more detailed view of customer segmentation, keep in mind that there is no one solution fits all scenario.
Benefits of a segmentation model?
Before jumping in and spending hours on building your segmentation model let’s examine the benefits. For one, it will allow you to differentiate your customers and therefore create different marketing strategies for different groups. For example, you can have a high value group to whom you may want to email on a weekly basis especially any new product offerings vs. a low value group which you can get away from only emailing once every quarter without harming your result. Another way to segment based on purchase history is by product categories or types and therefore speaking to different customers differently. For instance, if you are selling golf and tennis products, you may want to split your customers into three categories those who only play tennis, golf or both.
Having clear segments can help establish a target market that you wish to go after. You may realize that most of your revenue is coming from a small group of people that all have attributes that are the same. Therefore, you can now target similar people who look like your top customers to acquire more customers that will bring in more value.
All of the above will ultimately increase your revenue and if done right can help streamline your marketing and increase your marketing efficiency.
So now that you realize the benefits of the segmentation model let’s figure out the different ways of doing it.
Different types of segmentation models:
- Looking at different elements for multiple segments by customer data: The most common way that customers are segmenting their customers is based on different elements including:
- Geographic – where they are located. You can split your customer list by state, city, urban vs. rural and by countries or continents.
- Demographics – Using some demographic characteristics you can break your audience by age, race, household income, gender, religion, education, their family situation and what they do. If you are lacking this data in your CRM, you can use third party tools to append it. You would use this information to establish the different types of customers you have.
- Psychographic – Psychographic looks at personality traits, interests, lifestyles and so on. It is not as easy to capture it as demographics and are often guessed. You may be able to derive your own psychographic attributes based on the type of products that they purchase. If they purchase golf equipment on a yearly basis chances are they are golfers, if they purchase travel equipment they are travelers and so on.
- Looking at customer’s behavior: Instead of identifying who the customer is you can define customers based on their behavior. What they buy, how often they purchase from you, how many emails they open, how much they spent, do they spend 2 minutes to make a decision on adding an item to the cart or 3 hours, and so on.
You can use RFM model which looks at recency, frequency and monitory value to identify your best customers vs. worst customers. You can also use purchase history you can categorize them in different ways by product types or categories.
- Advanced Segmentation Model: Using cluster analysis you can put together customer’s behavior along with their demographic, geography and psychographic elements to establish clear segments. This model works best for those that have a lot of data and can actually do an advanced segmentation model looking at all different ways to segment their customers.
How to choose the best segmentation model for you?
While you can slice and dice your data differently and all those options may make it hard for you to get started here is a simple checklist that can help you:
- What is your objective? Having a clear objective can really help streamline your segmentation. For example, are you looking to identify a group of customers to upsell a specific product? Expand into a new region? Identify different target groups? All those questions and more can help you identify if you want to focus on a specific type of segmentation such as geographic segmentation, email marketing segmentation based on historical email performance and so on.
- Do you have the data and is it useable? Now that you have defined your problem, next step is to pull all your data from your CRM and see if what you are planning to do is available. You may need to acquire missing data, move on to a different segmentation model if you cannot get the data or figure a way to acquire the data going forward. If you do have the data is the format usable in other words standardized. For example, if you have age information because you have their date of birth you may need to extract the year and bucket that information into a specific age group.
- Establish your metric. Now that your attribute data is available and is in the right format you need to establish the metric that you will use for segmentation it could be average order size, lifetime value of the customer, RFM for RFM model and so on.
Now that you have chosen the right model for you and have all the data ready you can continue on to the analysis part.
Simple Approach: You can do a simple analysis where you divide each group into a few segments. For example, you may establish a segmentation for high value customers, low value customers or average. Therefore, looking at the lifetime value you would use an index compared to average and classify it into the three different groups.
This approach is perfect for those companies that have limited data, have a specific problem that you are trying to solve and for those that lack resources to do a more advanced approach.
Advanced Approach: For everyone else, there is an advanced approach which uses either predicative decision tree – model or cluster analysis to put together different attributes to come up with niche segments. Using machine learning packages in R or Python or statistical software packages such as SAS or SPSS you can load your data and produce statistical significant models. If you train your model correctly you would be able to identify which attributes have the biggest impact on your sales and look at multiple attributes at ones. While not all business have the capabilities in-house, hiring a segmentation analyst maybe a cheaper and faster solution.
Building a segmentation model is often the beginning of a longer journey. Once you have identified your groups it is now time to add the segments back into your CRM for quick identification for your marketing department. Next steps would include changing your marketing to match the different segments, it can include creating different email marketing campaigns, changing your digital marketing ads and so on. If done right, you should see an impact on your bottom line within a few months.