How to Measure Customer Loyalty: Metrics Beyond NPS

Why one number is not enough

Most owners who ask how to measure customer loyalty start with a single survey question: would you recommend us? The trouble is that people say one thing and do another. A customer may praise you on Instagram and still buy from the shop near their home. Another may complain about delivery and reorder every week.

So loyalty is measured on two sides. Behavioral metrics show what customers actually do. Survey metrics show what they think and are willing to tell others. The first tells you what is happening; the second explains why. Relying on just one gives you half the picture.

How to measure customer loyalty by behavior

This data usually already sits in your CRM, point-of-sale system or loyalty program, so no survey is needed. What matters is exporting it regularly and calculating it the same way each time.

• Repeat purchase rate: the share of customers who came back within a chosen period.
• Retention by cohort: take customers who first bought in the same month and see how many are still active later.
• Purchase frequency and interval: a growing gap between orders for one customer is an early sign they may be leaving.
• Share of wallet: how much of a customer's spending in your category goes to you rather than to a competitor.
• Length of relationship: especially useful for clinics, subscriptions and B2B.

Survey metrics: what customers think

NPS asks for a score from 0 to 10 on how likely someone is to recommend you. Respondents fall into promoters, passives and detractors, and the score is the share of promoters minus the share of detractors. It is simple and easy to compare over time, but it measures intention rather than action and does not explain itself.

CSAT measures satisfaction with a specific moment, such as a purchase, a delivery or a call with support. CES asks how easy it was to solve a problem. A customer who had to message you on Telegram three times to get a refund may still give a decent overall score, and then quietly choose another seller next time.

Always add an open question: why did you give this score? The answers are where concrete actions come from.

Referrals: the most honest proof

Willingness to recommend is not the same as an actual recommendation. Ask new customers how they found you and record the answer "a friend recommended" separately. You can ask it in an order form, in a Telegram bot or at the front desk.

Also watch the public trail: map reviews, Instagram comments, mentions in group chats. Volume matters, and so does tone.

Reading the metrics together

If retention falls while NPS holds steady, customers are probably still happy but leaving for outside reasons: relocation, seasonality, a competitor with a more convenient location. If NPS drops while purchases hold for now, treat it as an early warning.

Combining both views gives four groups. Truly loyal customers return and recommend. Habitual ones return but do not recommend, and may leave when an alternative appears. Hidden fans like you but buy little, so find out what stops them. The disloyal neither return nor recommend. Each group needs a different response.

Collecting data and avoiding mistakes

Keep surveys to one to three questions, sent right after a purchase or a support contact through a Telegram bot, an SMS link or email. Do not ask the same people too often, because survey fatigue distorts answers. Look at results by segment: new versus returning customers, branches, product lines. To keep sales, CRM and ad data from living in separate spreadsheets, many businesses bring them together in end-to-end analytics.

Common mistakes to avoid:

• Relying on NPS alone
• Not asking why a score was given
• Surveying only happy customers
• Comparing your NPS with other companies without considering industry and method
• Collecting answers and changing nothing, which irritates customers

A minimal toolkit to start

A small business can begin with three metrics: repeat purchase rate, NPS with a follow-up question, and the share of new customers who came through a recommendation. Together they show behavior, attitude and real word of mouth. Add more once these three are tracked consistently.

If your customer data is scattered across a till, Instagram, Telegram and spreadsheets, markdev can help set up end-to-end analytics and reporting where loyalty sits next to advertising and sales. Tell us about your task and we will suggest where to start.