How to identify frequent customers in a physical store?

How to identify frequent customers in a physical store?

Arturo A.

Digital Marketing Expert and AI Enthusiast

How do you identify frequent customers in a physical store? Industry thresholds, step-by-step RFM calculation, and automated win-back campaigns.

The most widespread piece of advice for identifying frequent customers is counting receipts: anyone who reaches three purchases enters the list and starts receiving promotions. That rule fails on two fronts. Three purchases do not mean the same thing at a daily-consumption coffee shop as they do at a special-occasion restaurant, and a loose count ignores the date, the amount, and the probability of that person returning.

A brick-and-mortar store does not need a loyalty program to start measuring. It needs a receipt-level identifier, a time window, and a threshold tied to its specific industry. Everything else follows naturally: segments, campaigns, and a dollar figure that shows how much it costs to win back someone who is slipping away.

The uncomfortable truth is that the face the cashier recognizes is not always the frequent customer. Rewarding someone who was already going to return is giving money away; the customer who is drifting away almost never comes up in counter conversations. Learning how to identify frequent customers in a brick-and-mortar store starts with accepting that staff memory is not a tracking system.

The visit threshold that defines a frequent customer according to your industry

A customer is frequent when they exceed the typical frequency of their industry within a defined time window. Not when the cashier greets them by name, not when they reach a round number of purchases, and not when they are in the business's WhatsApp group. The threshold is calculated using your own transaction history, documented, and reviewed every quarter.

Your type of business sets the baseline: in a coffee shop where a large share of customers come back every week, someone who shows up once a month is not frequent; they are an occasional guest who happened to remember you.

Illustrative frequency thresholds by business type

Business type

Visits in 30 days to be considered frequent

Recency window (active)

Risk signal

Daily-consumption coffee shop

5 or more

0–14 days

15 days without visiting

Neighborhood bakery

4 or more

0–21 days

22 days without visiting

Taqueria or quick service

4 or more

0–30 days

31–60 days without visiting

Special-occasion restaurant

1–2

0–60 days

61 days without visiting

Grocery or convenience store

6 or more

0–10 days

11 days without visiting

None of the values in the table are market statistics. They are operational starting points, measured in days, adjustable to each location's history. The logic of the purchasing cycle is confirmed by consumer behavior itself: in a study published in Salud Pública de México using data from Mexico's 2006 National Health and Nutrition Survey (ENSANUT), 60% of adults consumed a street-vendor snack or beverage at least once a month, while 94% had dinner at full-service restaurants or fixed-price meal diners less than once a month. The same habit, two different cycles. Applying the coffee shop table to a full-service restaurant produces a database full of false frequent customers.

How to calibrate the threshold with your own history

Export the receipts from the last 90 days with identified customers, group them by individual, and sort from highest to lowest number of purchases. Take the median of that list and use it as your baseline: anyone above your own operation's median is frequent for your store, even if they don't reach the number in the table.

The exercise takes an afternoon and corrects two biases at once. If your median is 2.4 receipts in 90 days, a threshold of three monthly purchases would leave you with no customers to reward. If your median is 11, almost everyone already clears a threshold of four, and the label loses its value.

The five operational segments

  • New: a single receipt within the table's window. There is no habit yet, just a trial.

  • Returning: two or three purchases, below the industry threshold. They are close to becoming frequent and represent the segment with the most room for improvement.

  • Frequent: exceeds the threshold and remains within their recency window. This is the customer to recognize, not the one to discount.

  • Inactive: their recency window has passed without a return, but they haven't exceeded twice that window.

  • Reactivatable: they stopped visiting longer ago than double their window, yet have responded to a message at least once.

The universal three-purchase rule erases these differences. Two people can spend the same amount in a month and be at opposite ends of the spectrum: the first visits four times and spends $400 MXN; the second makes a single $400 MXN purchase. In a sales report, they look identical. In a frequency model, one is the lifeblood of the business and the other barely paid a visit. Any customer segmentation table that fails to distinguish between those two trajectories is good for sending emails, not for making business decisions.

How to capture the identifier at checkout and calculate frequency, recency, and value

Frequency is not estimated; it is divided. The entire model fits into three numbers per person: how many receipts they have in a period, how many days have passed since their last purchase, and how much money they spent. With those three, you decide who to reward, who to wake up, and who to leave alone.

  1. Define a unique identifier and normalize it. A 10-digit phone number is the most practical key in Mexico: almost every customer provides it verbally and almost no one changes it. Email or an app QR code work as backups. Save it without spaces, hyphens, or the +52 country code, because the same number formatted three different ways creates three distinct customer profiles. If you have never built such a record, this customer database guide outlines the essential fields.

  2. Capture six fields per receipt. Date and time, store location, amount in MXN, items purchased, cashier, and customer identifier. Without the cashier, you cannot audit the record; without the store location, you cannot compare stores or consolidate customer histories.

  3. Ask for the information at checkout, not afterward. Use a brief phrase and highlight the benefit first: “Could I get your phone number to log today’s visit?” Asking for the detail via WhatsApp the following day cuts the response rate in half.

  4. Frequency = (period receipts ÷ period days) × 30. This 30-day normalization allows you to compare a customer who has been visiting for three months with one who has been visiting for three weeks.

  5. Recency = days elapsed since the last receipt. Calculate this against today’s date, not the reporting cutoff. A database that ages an entire month ends up sending promotions to people who have already returned.

  6. Value = total spend over the last 90 days and average ticket (spend ÷ number of receipts). Both matter. Total spend indicates the relationship's value; average spend indicates how much each visit counts.

  7. Assign an RFM score from 1 to 5. Rank your customers by frequency, recency, and value, and divide them into quintiles. The result is a three-digit score used to prioritize: a 5-5-4 is not handled the same way as a 2-5-5.

Baseline rule: an identifier without a date cannot measure frequency, and a date without an amount cannot measure value. All six fields must be captured together, or the model collapses.

The complete process, from checkout capture to scoring, is summarized in this infographic.

Infografía que explica los cinco pasos para capturar datos de clientes y calcular el modelo RFM en ventas.

Worked example with two coffee shop customers

One customer has 9 receipts in 90 days, her last visit was 4 days ago, and she has accumulated $1,485 MXN. Her frequency is (9 ÷ 90) × 30 = 3.0 visits per month, her average ticket is $165 MXN, and her recency is 4 days. She is well within her window, returns every ten days, and spends consistently.

The second customer has 2 receipts in 90 days, his last visit was 52 days ago, and he has accumulated $1,600 MXN. His frequency is 0.67 visits per month, his average ticket is $800 MXN, and his recency already puts him outside the window. He spends more in absolute terms, but his relationship hangs on just two visits, and he hasn't shown up in nearly two months.

In a sales report, the second looks like the best customer of the two-month period. In the model, he is an at-risk high-ticket customer with a real likelihood of never returning. The correct action is not a reward coupon; it's a win-back outreach.

Choosing where that data lives is the next decision. This guide to choosing the right CRM for a small business in Mexico organizes the criteria that matter to a food business, from checkout capture to the reporting that the owner will actually review.

Tactics to get customers to identify themselves and ensure the team always logs them

Customer databases don't fall short due to a lack of customers; they break down during peak hours. At 2:00 PM, there are six people in line, the cashier processes payments in eleven seconds, and asking for a phone number feels like a luxury. At the end of the day, the report shows 40 receipts and only 6 identified customers.

The consequence is quiet: the database grows only with customers who visit during slow hours, who are rarely the most frequent ones. Later, campaigns are designed around a skewed sample, leading the owner to conclude that messaging doesn't work.

Which tactic withstands peak hours

Tactic

Extra checkout seconds

Customer friction

Data quality

App QR code scanned at checkout

3–5

Low, the customer already has it ready

High, unique identifier

Phone number dictated to cashier

10–15

Medium, spoken aloud in public

Medium, risk of typos

Physical punch card

5–8

Low

Low, no contact details or digital history

WhatsApp signup via counter link

15–20

Medium, requires a smartphone

High if the number is confirmed

Email written on the receipt

20–30

High, slows down the line

Medium, useful for email, not for direct messaging

The table speaks for itself. Any tactic that takes more than ten seconds requires a clear counter incentive or a cashier who executes it automatically.

The one-sentence script and the limits of consent

Counter rule: offer the benefit before asking for the information, in a single sentence, and never ask twice.

At a coffee shop, this works: “May I have your phone number to track your visits? Your fifth one is on us.” At a taqueria, the tone shifts: “Can I get your number to log today’s purchase? On your seventh, we’ll give you a discount.” The script is finalized, rehearsed, and taped next to the register.

Trust is managed by keeping the privacy policy in plain view, explaining what the phone number is used for, and making opting out as simple as replying with a single word. An academic study presented in 2025 on consumers and smart technologies found that for 47.6% of surveyed consumers, smart in-store technology positively influences their decision to buy again, especially when it includes recommendations based on their purchase history and personalized promotions, while 28.4% express concern about the use of their personal data, according to the paper presented at LACCEI. Both statistics share the same checkout line: personalization adds value, while mistrust erodes it.

Ilustración con tácticas para solicitar la identificación del cliente y realizar el registro oficial de sus datos.

Do not ask for physical addresses, full dates of birth, or official IDs. Frequency, last purchase date, ticket size, and preferences are all you need to segment effectively.

The primary internal metric is simple: the percentage of receipts with an identified customer, reviewed weekly by cashier and store location. A reasonable initial goal is 35% in month one and 60% by month three. Furthermore, a database of 800 verified phone numbers performs better than 4,000 with fake numbers or single-letter entries: sending a campaign to the wrong customer costs just as much as sending it to the right one.

When visit counts lie and how to fix it

Frequency reports consistently fail for the same reason: they measure receipts, not people. A customer using two phone numbers, a family order logged as a single visit, or a store location maintaining a separate transaction history skews the median—and with it, the entire model.

Common pitfalls when measuring frequency and their fixes

Symptom in your data

Probable cause

Fix

A regular appears as a new customer

Two phone numbers or two profiles

Normalize to 10 digits and merge histories

Many customers with only one visit per month

Group receipts or family orders

Track guest counts or units per receipt

The same customer is new at every location

Separate transaction records per store

Consolidate profiles across the entire business

Digital buyers appear as occasional shoppers

WhatsApp orders handled outside the POS

Link the order to the same identifier

Registration drops during the afternoon shift

Cashiers skipping the capture step

Audit the percentage of identified receipts per cashier

High-ticket buyers look like the best customers

Looking only at transaction totals

Separate frequency from spend using RFM

The omnichannel scenario

Checkout counts fall short by design. In Mexico, over 66% of shoppers researched online before buying in-store in 2025, and almost 63% always or mostly compared products in a physical store and then bought them online, according to Statista data on omnichannel shoppers in Mexico. A consumer who inquires via WhatsApp, compares options at the counter, and buys through an app might look occasional across all three channels while actually being your most profitable customer of the year.

The operational fix is inexpensive: store the acquisition channel as a fixed field upon initial capture. If an in-store sale originated from a WhatsApp conversation, tag that receipt accordingly. This enables proper attribution and, crucially, prevents sending a reactivation discount to someone who already orders weekly through another channel.

Monthly cleanup routine

On the first Monday of every month, check three things before launching any campaign: duplicates from mistyped phone numbers, numbers with fewer than ten digits, and customers with two active profiles. Next, recalculate your industry's frequency median, as a threshold established six months ago might already be outdated. An unmaintained database depreciates faster than physical inventory.

Win-back campaigns that trigger automatically based on each customer's recency

The golden rule of win-back campaigns is uncomfortable: never offer a discount to someone who would return out of habit anyway. Every dollar discounted for an active frequent customer is sacrificed margin, whereas that same dollar, directed at an inactive segment, wins back an entire visit.

That is why counting how many customers registered reveals very little; what matters is how many returned and how long it took them.

As an operational example, consider these working tiers: active if they purchased within the last 30 days, at risk between 31 and 60 days, and lost past 60 days. These are illustrative ranges that each business adapts to its own cycle.

Four plays ranked by cost

Active frequent, no discount. The most cost-effective move is recognition: a visit-based dynamic that accrues and offers a free product on the seventh visit. It costs less than a percentage discount and reinforces the habit. In Swirvle, the rewards program lets you choose the reward structure based on your industry, whether that means visits, points accumulation, or purchases. One documented case describes it this way: “Every six visits, on the seventh we give them a discount. People feel taken care of,” notes Jesús V., from Pollos Vidaurri.

Ascending returning customer, with a measured nudge. The goal is to move from two to four visits per month. A reward on the third purchase of the month works better than an open coupon because it sets a deadline without giving away margin to someone who was already planning to stop by.

Inactive for 31 to 60 days, win-back coupon. A $50 MXN coupon on purchases over $200 MXN protects the ticket size and caps the discount cost at 25% or less. This is where automation comes in: if a customer doesn't visit for 30 days, Swirvle can send them an incentive to return without anyone having to trigger it manually.

Lost, a single message and silence. Beyond 60 days, send a single reactivation attempt, with no follow-up reminder sequence. Pestering an audience that has stopped responding burns out the channel and increases opt-outs.

Loyalty is measured by the visit that happens, not the message that gets sent. A program that only pushes promotions without tracking visits is just a discount newsletter.

Choosing reward structures by industry prevents the most common mistake: visits for coffee shops and bakeries, where consumption is high-frequency with low ticket sizes; points for grocery stores, where selections are broad and customers compare items; purchases for special-occasion dining, where visits are infrequent and ticket amounts are high.

From receipt to transaction history without manual entry

None of this works if checkout and customer tracking operate in silos. A retail POS system that recognizes your customers exists specifically to bridge that gap: every receipt is linked to a customer, and win-back messages trigger on their own. When automation integrates with retail marketing, business owners spend less time deciding who to message and more time assessing what worked.

The entire funnel, from registration to recovery, is outlined in this infographic.

Infografía sobre cómo automatizar campañas de retorno para recuperar clientes inactivos y maximizar ventas.

How to measure campaign success

Track three metrics per campaign: Sales attributed to the coupon, counted solely from receipts bearing the promo code. Incremental visits, benchmarked against a lookalike group of customers who did not receive the message. Incentive cost per recovered visit, calculated by dividing total discounts redeemed by the visits that occurred. If that last number exceeds the gross margin of an average visit, the campaign is acquiring visits at a loss.

Launch your customer database this week and measure the first month

A customer tracking system can be implemented in seven days if kept focused. The common pitfall is buying software first and figuring out the process later; the correct approach is the reverse.

  • Day 1. Choose your identifier (a 10-digit phone number) and set your industry's frequency threshold using your last 90 days' median.

  • Day 2. Train the team using the one-sentence script and set your target percentage for identified receipts during week one.

  • Days 3 to 7. Capture customer data with every transaction and review cashier capture rates nightly. By day seven, you will have your performance baseline.

To avoid relying on a manual notebook, your system needs three features: customer capture built right into the checkout workflow, every receipt automatically added to customer history, and win-back messages that trigger autonomously. An all-in-one CRM, loyalty, and POS platform like Swirvle provides all three: every sale is logged to customer profiles, re-engagement campaigns run automatically when a patron lapses, and a central dashboard monitors capture rates across cashiers and locations. The POS starts at $1,199 MXN per month per branch, VAT included; WhatsApp messaging requires the Premium plan, and the software requires an active internet connection.

At the end of month one, review two metrics and nothing else: how many frequent customers you gained, and how many inactive customers returned. Those two figures determine whether your process is actively driving revenue or simply hoarding phone numbers. Schedule a 15-minute demo to evaluate your store's data and see what threshold makes sense for your business.

Identifying frequent customers stops being guesswork once every receipt is tied to a person, with a date, location, and purchase amount. Swirvle brings together CRM, loyalty programs, and POS into a single operation, ensuring data is collected at checkout and follow-ups happen automatically when someone stops visiting. If you want to see how your own data performs with industry-specific thresholds and recency-based campaigns, visit Swirvle and schedule a 15-minute demo with your store's transaction history ready to go.

Try Swirvle for free

Cardless · loyalty program

Start your free trial