Customer Segmentation: Groups You Can Use
Customer segmentation divides customers into groups that share characteristics, such as demographics or purchasing behavior, so communication and offers can be tailored to those groups, according to Qualtrics’ definition of customer segmentation. Each group is a customer segment.
Qualtrics distinguishes customer segmentation from market segmentation: customer segmentation examines a business’s customers, while market segmentation considers the broader marketplace. Define that population before comparing groups.
Segmentation has uses beyond selecting a marketing audience. Bain’s customer segmentation guide describes its role in product development, product features, service options, distribution, pricing, and marketing. The intended use determines which customer differences matter to the analysis.
Main types of customer segmentation
The following dimensions describe different ways to group customers. Select the ones that relate to the decision being made.
| Type | Basis for grouping customers |
|---|---|
| Demographic | Individual characteristics such as age, income, education, and household composition. |
| Geographic | Location within geographic boundaries, such as countries, regions, or cities. |
| Psychographic | Shared values, interests, opinions, beliefs, and lifestyles. |
| Behavioral | Purchase activity, product usage, engagement, and loyalty-related behavior. |
| Needs-based | Similar customer needs or unmet requirements. |
| Value-based | Economic differences assessed through the revenue and cost impacts of serving a group. |
| Firmographic | Organizational attributes such as company size and employee count, when the customer is an organization. |
These dimensions can be combined: Qualtrics’ segmentation guidance recommends selecting a combination appropriate to the objective.
For economic segmentation, distinguish spending from profitability. Bain’s approach includes both revenue and the cost of serving customers; record which measure a value-based segment uses. For psychographic or needs research, use direct responses about preferences and priorities: Qualtrics recommends collecting psychographic information from customers themselves.
Methods for creating customer segments
Rules-based segmentation
Rules-based segmentation assigns customers through explicit conditions; Shopify’s customer segmentation system combines filter names, operators, and values, then automatically updates membership as customers meet or stop meeting the criteria.
For a rules-based approach, document each field, condition, and observation period. State how multiple conditions combine and how missing information is handled. Use names that describe the recorded characteristic, and retain the underlying rule alongside the name.
RFM analysis
Shopify’s RFM documentation defines the three measures used to describe purchasing patterns:
- Recency: time since the most recent purchase.
- Frequency: number of orders.
- Monetary value: total amount spent.
Use RFM for questions about purchasing history. Add customer research when the objective also requires information about needs or preferences.
Clustering
The scikit-learn clustering documentation describes algorithms for grouping unlabeled records using selected input variables. Its k-means method requires a specified number of clusters and minimizes the sum of squared distances between records and their cluster centers.
Interpret each cluster in relation to the variables and algorithm used to create it. Inspect its distinguishing characteristics before assigning a business meaning or action. Document the inputs and method, then check whether new customer records can be assigned through the same process. Choose the method according to the available data and intended use.
Documented examples of customer segments
Shopify’s RFM customer groups provide concrete examples of segment definitions and associated engagement recommendations:
| Shopify group | Customer characteristics | Suggested engagement in Shopify’s guide |
|---|---|---|
| New | Very recent purchases, few orders, and low spending. | Provide onboarding support and begin building the relationship. |
| Champions | Very recent purchases, many orders, and high spending. | Provide early product access or introduce a referral program. |
| At risk | No recent purchases, with a strong history of orders and spending. | Use tailored communication and an offer to encourage a return. |
Use these source-specific definitions with their published assignment rules, and evaluate the suggested actions against the chosen business outcome.
How to build customer segments
1. Define the objective and population
Name the decision the segmentation will support: message selection, product development, service allocation, or another specific use. Specify which customer records are included, the unit being grouped, and the period being examined. Keep customer, household, and organizational records distinct when they represent different units of analysis.
Choose the outcome to measure at this stage. A campaign objective needs a campaign outcome; a service objective needs a service measure. Record the treatments or changes that are feasible before deciding how finely to divide the population.
2. Collect relevant data
Use data that relates to the objective. Qualtrics’ customer segmentation guide describes direct information from customer surveys and indirect streams such as analytics and frontline feedback.
Select transaction records for purchasing questions, usage records for activity questions, and direct responses for stated needs or preferences. Record where each variable comes from and when it becomes available. Use only information available at the moment a customer must be assigned.
3. Prepare consistent records
Check customer identifiers, duplicate records, incomplete fields, and date formats. Define how refunds and canceled orders affect purchase measures. Set an observation window for activity and spending, and specify the treatment of customers without enough history.
Keep an explicit unassigned category when required information is unavailable. Record missing information separately from a measured absence of activity. These are recommended data preparation steps for making assignment rules reproducible.
4. Create and check candidate groups
Apply rules, RFM, or a clustering method using the selected inputs. Examine group size, distinguishing characteristics, missing information, and overlapping membership. Check that assignment uses the intended population and period.
Assess each group against the criteria below. Retain a distinction when it supports a separate decision, treatment, or research question. Combine groups when their separate definitions do not serve the stated objective.
5. Document the action and assignment rule
For each retained segment, record its name, inclusion and exclusion criteria, data sources, observation window, assignment method, and refresh schedule. Add the planned action, its owner, and the outcome to measure.
Specify how overlapping segments affect action selection. For customer records awaiting assignment, define a default treatment. Keep a dated version of the definition so that later results can be interpreted against the rule used at the time.
What makes a customer segment useful?
Qualtrics identifies four criteria for effective segmentation: measurable, accessible, substantial, and actionable. Applied to customer groups, these mean:
- Measurable: the relevant purchasing potential can be estimated.
- Accessible: the customers can be reached through an appropriate channel.
- Substantial: the group has the ability to purchase the offering.
- Actionable: the distinction supports a different response to the offering.
Use these criteria to review the proposed groups before activation. Add a practical delivery check: identify the resources needed to provide the planned message, product, or service to each group.
Measure results and update the segmentation
Bain recommends measuring segment performance and adjusting the approach over time. Choose measures connected to the action: conversion or repeat purchasing for a campaign, continued purchasing for retention, and revenue alongside service costs for economic decisions.
Where a controlled test is feasible, compare the existing treatment with the proposed treatment within a predefined segment. Optimizely describes A/B testing as random assignment to two variants followed by analysis against a conversion goal. Apply the test to the selected group and evaluate the tailored page or experience against that goal.
Define the outcome and evaluation period before starting. Check that the segment has enough traffic for the intended test; Optimizely’s guidance on segmented experiments warns against dividing tests into excessively small audiences.
Review assignment coverage, changes in group size, and missing data alongside the business outcome. Preserve the segment definition used for each evaluation period. Revise the rule when the relevant customer differences, available data, or intended action change.
Distinguish membership updates from changes to the model. Shopify’s dynamic segments add or remove customers as their data meets the existing conditions. Changing those conditions is a separate revision; record its effective date and check how it affects assignment.
Frequently asked questions
How many customer segments should a business have?
Choose the number that preserves the differences needed for the stated objective and the actions that can be delivered. Qualtrics notes that segmentation approaches vary with business objectives and circumstances. Use a separate group when its definition supports a distinct decision or research question.
What is the difference between customer segmentation and personas?
A segment is a group of customers sharing defined characteristics; a persona is a representative description of that group, as Qualtrics explains in its discussion of personas and customer segments.
Are demographics enough for customer segmentation?
Demographics provide descriptive attributes such as age, income, and education. Use them when those characteristics answer the question being studied. For a broader view, Qualtrics describes combining demographic, transactional, behavioral, and psychographic data. Select the inputs needed for the objective.
How often should customer segments be updated?
Match assignment updates to the rate at which relevant data changes and decisions are made. Dynamic systems such as Shopify’s customer segments update membership when customers meet or stop meeting the rules. Review the definitions separately when the objective or customer conditions change.
Is customer segmentation the same as personalization?
Segmentation defines customer groups, while personalization tailors communication or experiences; Shopify’s segmentation documentation describes combining the two through targeted campaigns with personalized messaging. A segment supplies the audience for that tailored communication.