How to Identify Loyal Customers: Separate Behavior, Preference, and Recommendation Intent

Published:
Last Updated:
Category: Marketing Glossary, CRM, LTV & Customer Management
Authors: Shusaku Yosa
Identifying loyal customers requires more than a purchase ranking. Look at behavior and investigate why someone chooses the business. High spenders, people with strong recommendation intent, and people who have actually referred others can overlap, but they are not the same group. This guide explains how to find candidates while keeping unknowns visible.
1. Decide why you need to identify them
A loyal user is commonly understood as someone with preference or support for a service and an ongoing relationship with it. There is no universal purchase count or spending threshold that identifies loyalty across every category. Supporting continued use, gathering improvement ideas, and inviting referrals require different information.
For an improvement interview, selecting only frequent users can exclude the people who encountered obstacles and stopped. State the purpose and target population first. Creating a group labeled loyal should not become the objective itself. Confirm whether people want contact or participation rather than assuming the label grants permission.
2. Choose a period that fits the category
A weekly service and a product replaced every few years have different usage opportunities. A lack of recent repurchase in the second category does not necessarily indicate weak preference. Automatic purchasing or contractual renewal also does not establish an active choice.
Define the starting point, observation window, and reasonable opportunity to use or buy. Comparing new and established customers by lifetime spending mixes relationship duration into the result. Align tenure where appropriate and leave incomplete observation windows undecided. Consider pauses, returns, and other information that a simple purchase count can miss.
3. Find behavioral candidates in the records
RFM organizes recency, frequency, and monetary value. It can help locate behavioral patterns, but it does not directly measure attachment. For a usage-based service, define relevant completed tasks or continuity measures instead of substituting activity volume for value.
Observation | Possible source | What it establishes | Question remaining |
|---|---|---|---|
Recent purchase or use | Transaction or activity record | Timing of a defined action | Was another opportunity expected? |
Frequency or continuity | Records within a stated window | Repeated behavior | Preference or contractual constraint? |
Spending | Purchases and refunds | Spending within scope | Profit and attachment remain separate |
Choice reasons | A customer's answer | The stated explanation | Nonrespondents' reasons are unknown |
Referral | A verified referral record | A defined referral action | Intent and recipient outcomes are separate |
Document each selection rule and its period. A commercially valuable customer is not automatically willing to join every research or promotion activity. Check the appropriateness and burden of the invitation separately.
4. Ask about preference and recommendation intent
Use open questions about what makes the service useful and what remains difficult compared with alternatives. Avoid prompting praise, and make room for improvement requests. An unanswered question means the reason is unknown, not that preference is low.
NPS classifies respondents choosing nine or ten as promoters, but the answer is not a record of an actual referral. Refer to Bain's official explanation and keep intent, advocacy activity, and the recipient's use in separate fields. Someone may be willing to recommend yet have no suitable occasion to do so.
5. Compare four fictional profiles
These invented profiles illustrate questions to investigate. They are not a universal classification system or a ranking of customer worth.
Person | Observed behavior | Stated explanation | Provisional interpretation |
|---|---|---|---|
A | Monthly use | Changing the contract is difficult | Repetition observed; active preference unconfirmed |
B | A purchase every two years | Chooses this brand when needed | Possible strong preference; check the category cycle |
C | Continued use; no referral record | Recommendation score of nine | Intent observed; actual referral unconfirmed |
D | High spending | No answer | Commercial behavior observed; reason unknown |
Use these distinctions to choose a relevant response. Ask A about constraints and inconvenience. Contact B at an appropriate purchasing occasion. If inviting C to refer, ask whether participation is welcome rather than assuming consent from the survey answer.
Keep the date and source of each observation. If records and answers appear inconsistent, investigate timing, recent experiences, or changed alternatives instead of deleting inconvenient evidence. Four examples cannot establish the proportions or future behavior of the full customer base.
Treat selection thresholds as local rules to test. In a fictional recurring service, three uses in the last 90 days could identify candidates, but the threshold would not prove attachment. Check whether it misses supportive users or includes people staying because of contractual constraints. Keep a provisional, awaiting-confirmation category when information is incomplete.
6. Review the group and choose useful action
Set a review point rather than assigning a permanent label. Reflect changes in purchase cycles, unresolved issues, and communication preferences. Avoid continuing special treatment that a customer no longer wants.
Next steps might remove a first-use obstacle, confirm an important use case, or offer a voluntary opportunity to contribute. Evaluate whether the response helped, not merely how many candidates were found. Records identify behavior; conversations add reasons; explicit unknowns prevent stronger claims than the evidence supports.
Related reading




