Customer segmentation is a means by which you group customers into an identifiable category upon which you as a company can analyze for among other activities strategic planning, market campaigns and architecting customer experiences.
Factors that could be used in customer segmentation include
- Firmographic could include # employees, industry, etc.
- Demographic could include age, marital status, location, income
I like behavioral rather than segmentation based on attributes or a derived factor. Behavioral is based on something the customer DID, rather than an attribute externally assigned to the customer. Some good sources of behavior-based attributes include.
- Transactional history (soon to come my post about churn management and customer accounting). Within transactional history, a popular method of segmentation is using cohorts based on time of transaction. Other metrics include segmenting customers on recency, frequency and monetary value of the transaction.
- Social profile, which included connections, networks and affiliations both on- and offline.
- Sentiment, which is associated with attitude, but more specific to sentiment analysis as online (watch out for future blog on this).
Factors used for customer segmentation need to be SMART.
- Specific – Clearly defined scope
- Measurable & Manageable. You need to be able to measure it and collect the data reliably.
- Should be based on a hypothesis that you can prove or disprove with data or information available to YOU at a reasonable cost
- Actionable within your organization – somebody should own the decision and thus the ability to prove/disprove a hypothesis
- Relevant to your goal: Current drives + future accelerators = Value bringing you towards your goal
- Trending – The data and information should be available over a significant window of time & available at the necessary intervals
Rule based and analytics driven
- Rule-based segmentation is based on criteria such as Boolean logic or thresholds and is usually two dimensional
- Analytics driven segmentation such as k-means clustering is a data-driven means of determining customers who are similar along specific variables. Because these clusters are based on similarity and statistical significance, they are helpful in predictive analytics.