A discount card answers only one question — what percentage to take off this receipt. It does not remember who is standing at the checkout, how much they bought over the year or whether they will come back at all. A loyalty program solves a different problem: recognize a regular customer, calculate their points under clear rules and give them a reason to come again. And for the first time the company sees not a stream of receipts but the people behind them.
Loyalty program — the rules by which a company works with its regular customers, and the software that carries those rules out by itself. A customer enrols once, and from then on every purchase they make is recognized, enters their history and earns them points. The company gets something it never had before: a list of its buyers and an understanding of which of them come back.
The difference is clear from a single example. A 10% discount on a plastic card is a one-off concession at the moment of payment: the money is gone and the customer is forgotten. The same 10% given as points is a reason to come a second time, because they can only be spent with you. The first reduces today's revenue; the second brings the customer back in a fortnight.
A points program — is the most common loyalty mechanic but not the only one. Tiers, personal offers, promotions for a specific group of buyers and purchase history all work from the same database and the same data. Almost everyone starts with points: they need no explaining to the customer and are calculated automatically.
An example. A customer left their phone number at the checkout — the customer card is created. They spend 3,200 som, the system credits them 160 points and records the receipt in their history. Two weeks later they come back, give the same number and pay for part of their purchase with points. By the fourth visit the total of their purchases moves them to the next tier and the earning rate goes up. The cashier calculated none of this by hand.
Loyalty program automation begins where the answers to three questions no longer fit in the administrator's head: who this customer is, how many points they have right now, and on what terms they earned them.
This is a circle, not a straight line: the seventh step brings the customer back to the second, and with every turn the system knows more about them. The point of a loyalty program is to close that circle at least once: a customer who has used their points comes back noticeably more often than one who simply accumulated them and forgot.
It does not create demand and does not retain a customer in place of the product and the service. It removes the manual work around a purchase: it recognizes the person, calculates points under the rules that were set, remembers their history and prevents double crediting. The decision on what percentage to give back and who gets special terms stays with the company — but it is made from figures rather than from a feeling.
In just the same way, the system does not see the customer by itself: it knows exactly what reached it with the purchase. If half the receipts bypass the program, the reports will describe not the business but the part of it that got recorded.
Two things come first: a way of recognizing the customer at the checkout and one earning rule across all products. The reason is simple: while the customer is not recognized there is nobody to credit, and a dozen rules with exceptions cannot be explained either to the cashier or to the buyer. Tiers, promotions and personal offers are added later — once the database holds enough history to build them from.
What follows is not a feature list but six problems that companies set up a loyalty system for in the first place. Each one is stated the same way: what happens without software and what changes with it.
Without software a company knows its revenue but not its buyers: a receipt is anonymous, and whether the person came back or came for the first time is unknown. A program member is a record with a purchase history, so how many regular customers do we have stops being a matter of guesswork.
A discount reduces revenue immediately and promises nothing in return. A point is an obligation the customer can only realize with you and only on their next purchase. The money stays in the till today, and a reason to come back appears for the future.
While percentages and exceptions rest on the cashier's memory, one customer is credited more and another less, and disputes are settled by whoever is louder. The rules are set once in the panel and applied identically at every location and in the online store.
A customer leaving is not accompanied by anything: they simply stop appearing. The purchase history shows that someone who used to come every two weeks has not been in for two months — and that can be turned into an offer while they still remember your sign.
An identical mailing to the entire database produces unsubscribes and irritation. Purchases show who buys cat food and who buys children's clothing — and the offer goes to the people it could conceivably interest.
Without a program, the result of a promotion is judged from the month's total revenue, where its contribution is indistinguishable from the season. Here you can see how many members used it, how many points went out and whether those people came back afterwards.
The system is assembled from modules. Not every company needs all of them: a single-location coffee shop is fine with a points account and tiers, while a retail chain cannot manage without segments and exchange with the checkouts. The composition is determined by the task, but the modules are designed to fit together in advance rather than being bolted on later.
The customer card: contact details, enrolment date, consents, points balance, tier and the whole purchase history. This is the foundation of the program — everything else is calculated from it rather than maintained separately.
A way of recognizing the customer at the moment of purchase: phone number, QR code, virtual card, app or an account in the online store. The set is chosen per project.
The customer's balance, accruals, redemptions and the validity period of every batch of points. Every operation has a reason behind it — a receipt, a promotion or a manual adjustment with an author.
How much is earned and on what: a percentage of the total, a separate percentage by product category, a raised rate during a promotion. The rules are set in the panel rather than in the checkout's code.
The conditions for moving up and what a tier gives: earning rate, discount, access to specific offers. The tier is recalculated by the system rather than assigned by hand.
Time-limited mechanics: points for a purchase, a raised earning rate, a discount, a promo code. Every promotion has an audience, a trigger condition and a report.
Groups of customers by clear criteria: bought in the last month, have not been in for six months, buy a particular category, have accumulated more than a thousand points. A segment is an address for an offer.
All of the customer's receipts with contents, total, point of sale and points earned. Both personal offers and the resolution of a dispute at the checkout come from here.
Messages to the customer: points credited, the balance is about to expire, a new offer is available. The sending channels are chosen during rollout and connected through integrations.
Where the person sees their balance, tier and offers: a mobile app, a page in a browser or a section inside the company's existing app.
The company's workplace: customers, rules, points, tiers, promotions, segments, operation history and program indicators. It opens in a browser, with nothing to install.
The system's external interface: recognize a customer, request a balance, credit and redeem points, reverse an operation. It is how the checkout, the online store, CRM and accounting systems connect.
Identification — the moment a purchase stops being anonymous and is tied to a specific person. Without it there is no loyalty program at all: there is nobody to credit and nowhere to write the history. Everything else — rules, tiers, offers — is built on top of that step.
It works the same way whatever the method. The customer gives or shows their identifier, the checkout or the website passes it to the loyalty program, the program finds the card and returns an answer: who this is, how many points they have and which rules apply to them. It takes fractions of a second and happens before the receipt is closed.
There is one requirement for the method — it must not hold up the queue. Anything that demands extra actions from the cashier or has the customer hunting for a card in their bag stops being used within a month in practice. That is why the most workable option is usually the simplest: a phone number the person remembers anyway.
If the customer is not in the database, the purchase is not lost: the system can create a card against the same number right at the checkout and credit points for that receipt. Registration turns into one question to the buyer rather than a five-field form — they can fill in the rest themselves later.
What happens at the moment of identification:
The whole exchange takes fractions of a second and happens before the receipt is closed. If the loyalty program is unavailable at that moment, the checkout must close the receipt without points rather than stop: the sale matters more than the accrual, and what was missed can be credited later against the receipt number.
A person buys in a shop, then orders on the website, then comes to another location in the chain. If those are three separate records, the loyalty program does not work: the points sit in three places and the purchase history never adds up to a single picture.
That is why the card has a primary identifier — most often the phone number — and the other methods are linked to it. The QR code in the app and the account on the website then lead to one and the same customer, whichever location they pay at.
Taking part in the program is a voluntary act by the customer, and consent to data processing is recorded in their card together with the date. The same place holds their opt-out from mailings: someone who does not want messages must stop receiving them while remaining a member of the program.
The cashier does not calculate points and does not remember the rules. They perform one action — confirming the customer — while the amount to be earned and the balance available for redemption are filled into the receipt by the system. That alone removes half the conflicts: the figure was named by the software, not by an employee.
Some purchases will bypass the program anyway: the customer is in a hurry, has forgotten their number or does not want to take part. That is normal, and such receipts are not lost — they stay in the overall revenue, they are simply not tied to a person.
Something else matters: the share of identified receipts has to be visible from the first month. While it is low, any program report describes a minority of buyers, and the conclusion that loyalty does not work is drawn from incomplete data.
None of the options is mandatory and none is claimed as already built: the set is determined during discovery — from how the checkout works, whether the company has an app and who your buyers are. There can be several methods at once, but the customer card stays a single one.
A point — the company's internal unit, which a customer receives for a purchase and can only spend with that company. The rate is set by the company itself, and it is almost always kept simple: one point equals one som. Anything more complicated has to be explained to the buyer, and there is no time to explain at the checkout.
The whole mechanic fits into one line. The customer makes a purchase → the system works out which rules apply to them → credits the points → stores the balance → on the next purchase the customer can redeem part of what they accumulated. Nothing happens between purchases except the expiry of overdue batches.
What the system holds for a points account:
The limits are set by the company, not by the buyer. The usual setting is an upper cap on the share of a receipt that may be covered with points: no more than 30%, for example. Without it, purchases get paid for entirely with points, and the program starts working instead of revenue rather than for it.
Points usually have a limited life — that is what keeps them a reason to come rather than an endless liability for the company. Expiry is calculated by batch: every amount credited has its own date, and whatever expires soonest is spent first.
The main requirement for this mechanism is a warning. Points that vanish silently damage the attitude to the program more than having none at all: what the customer remembers is not the rule but the feeling of having been cheated.
A return reverses both sides of the operation: the points earned on the receipt are removed and the points redeemed are returned to the account. Otherwise a simple scheme appears — buy, collect the points, return the goods — and the program starts giving away the company's money.
A company can exclude certain categories from redemption — promotional goods, marked-down items, delivery, gift certificates. This is a setting rather than a hard rule of the system: the list of exclusions is determined during discovery.
The order of the lines is itself the rule the company sets during rollout. Here the accrual is calculated only on the part of the receipt the customer paid for with money: otherwise points are earned on points and the balance grows by itself. The redemption cap, the earning rate and the calculation base are all configurable — what matters is that they are explicit and identical at every location.
A tier is a way of telling customers apart without reviewing each one by hand. Someone who buys from you once a week and someone who dropped in once get different terms automatically: by total spend, by number of purchases or by another rule that has been set.
| A tier | How to reach it | Earning | What else it gives |
|---|---|---|---|
| start Basic | Immediately on enrolment | 3% | A points account, general promotions |
| step up Regular | Purchases from 30,000 som over six months | 5% | Offers for regulars, points last longer |
| step up VIP | Purchases from 100,000 som over six months | 7% | Personal terms, early access to promotions |
The names, thresholds and privileges here are illustrative — this is an example configuration, not our standard. There can be two tiers or five; they can be counted by total spend, by number of purchases or by visit frequency. One general rule applies: the customer has to understand what they need to do to reach the next tier, otherwise the tier does not motivate anything and simply exists.
The system recalculates it itself under the rule that was set — after every purchase or on a schedule. Manual assignment is also possible, but it stays a separate operation with an author and a reason: otherwise six months later nobody will be able to explain why this customer has special terms.
Most often the earning rate. Beyond that it depends on the company's aims: a discount, access to specific offers, a higher cap on paying with points, a longer life for them. The set of privileges is defined in the panel and can differ for every tier.
If the conditions are counted over a period, a tier can be lost — and that too is a rule that has to be settled in advance. A gentler option: the tier is kept, but the customer is warned that they are a certain amount short before the period ends.
A promotion is a temporary change to the usual rules: for a period, for certain customers or for certain products. There are many mechanics, but usually only two or three run at once: the more rules apply at the same time, the less likely the buyer is to understand what they were given and why.
A fixed number of points on top of the usual accrual: for a purchase above a certain amount, for a first order, for a visit after a long gap. A simple mechanic with a clear condition.
The same points account but at a higher percentage — for the duration of the promotion, for a product category or for a particular day of the week. It works where demand needs shifting rather than simply discounting.
A direct reduction of the receipt total for program members or for a specific tier. It is calculated by the system under the same rules as points, so the cashier does not have to remember the conditions.
A code the customer enters at the checkout or in the online store. It can be shared across a campaign or personal — in which case you can see exactly who used it and how many such people there were.
A promotion that does not apply to everyone: only for a new customer, only for those who have not been in for two months, only for VIPs. The group is defined by a segment rather than by a list assembled by hand.
Special terms on specific items or on a whole category. They are needed where the aim is not overall turnover but selling a particular group of goods: new arrivals, seasonal items, remaining stock.
The last line is the whole reason promotions are set up in a system rather than announced on a sheet of paper by the door. While the result is not measured, every next campaign is planned on impressions of the previous one.
An ordinary mailing is built the same way for everyone: one message goes to the whole database. Half the recipients never buy that category, some have not been in for a year, and someone came yesterday and has already bought everything. The response rate is low and the unsubscribes after every such mailing are constant.
Personalization here means something simple: instead of the same offer for everyone, the business uses what it already knows about a customer's purchases and activity and sends the offer to the people it might interest. This is not predicting the future but working with facts — what the person bought and when they were last with you.
An example. A customer bought cat food three times over the quarter and was last in six weeks ago. They get an offer in that category. A second customer, who only buys children's clothing, gets one in theirs. Both get it at the moment a purchase is already due on their usual cycle.
What can be used for selection:
All six criteria come from the purchase history already accumulated — there is no need to survey customers separately, fill in forms or keep lists in a spreadsheet. The longer the program runs, the more precise the selection: after three months habits are visible, after a year, seasonality.
Technically a personal offer is a promotion whose audience is defined by a condition rather than by a list. Bought category X, have not been in for 45 days, balance above 300 points is a segment the system reassembles itself at the moment of sending.
That is why a campaign does not go stale between preparation and launch: a customer who came in yesterday leaves the have not been in for a while group automatically and does not receive a message they should not.
Automatic recommendations of what this customer will buy and models based on artificial intelligence are not part of a basic loyalty program, and we do not promise them as a ready feature. Such mechanisms can be built separately — on the accumulated purchase history and for a specific task, if there is enough data for it.
The difference is fundamental: a segment based on actual purchases works from the first month and can be explained in words, while a forecast requires data, time and verification on real campaigns.
Message frequency is capped at program level: no more than so many per person per week, taking their opt-out from mailings into account. A customer who is written to every other day unsubscribes regardless of how precisely the offer was chosen.
A points account
Customer tiers
Promotions and offers
Program analytics
A customer needs somewhere to see their points. Without it the program exists only for the company: the buyer does not know their balance, does not remember the terms and learns about expiry after the fact. The question of how much I have accumulated reaches the company as a phone call, or does not reach it at all — along with the customer.
What the customer interface can show:
The form depends on the project. It can be a separate mobile app, a browser page behind a link, or a section inside the company's existing app — if there already is one, there is no reason to start another. The set of screens is determined during discovery rather than taken off the shelf.
A separate app is not justified for everyone: it has to be installed, updated and promoted, and a buyer only installs one for a noticeable benefit. A lighter option is a member page in the browser that opens from a link in a message and needs no installation.
The minimal scenario with no interface at all is a message with the balance after every purchase. It is less than a customer account, but it covers the main job: the customer knows how much they have accumulated.
Three things have to be readable at a glance: how many points there are, what date they are valid until and what has to be done to reach the next tier. Everything else — history, terms, an archive of offers — can sit deeper.
The order of the tiles is not accidental: first what can be used, then what can be lost. The expiring points tile brings people back to the shop more often than the other three combined — provided the customer saw it in advance rather than on the day of expiry. The figures are illustrative.
Management panel — the other half of the system: what the company sees while customers accumulate and spend points. Its job is not to show all the data but to make it possible to change the program's rules without a developer and to resolve a specific case when a customer at the checkout disagrees with something.
There is one main idea: the program's terms are set in an interface, not in code. A marketer changes the earning rate, launches a weekend promotion and edits a tier threshold themselves — otherwise every change turns into a development task and is put off for a month.
What the administrator manages:
The main screen in support work: contact details, enrolment date, tier, balance and every purchase as a list. It also shows where each point came from — enough to answer why was I credited less within a minute.
The search runs on the same data the customer identifies themselves with at the checkout: the number, the card code or the surname. There is no separate member directory maintained by hand in the system, and there should not be.
Crediting points over a complaint or removing ones given in error is a normal part of the work. But such an operation is stored with an author, a time and a reason, like every other: otherwise six months later the database holds accruals nobody can explain.
A cashier needs to recognize the customer and redeem points, a marketer needs to launch promotions, a director needs to see the reports. The separation is set by role rather than by individual rights for each person: otherwise a new employee's access is configured the same as Ivanov's, and nobody can say any more what exactly they have access to.
The third tile matters more than the first: while only some receipts go through with a card, every other figure describes that part rather than the business as a whole. That is where a program review starts a month after launch — raising the share of identified receipts is almost always cheaper than signing up new members. The figures are illustrative.
Reports are assembled from the same records that customers' purchases create — no separate data entry for analytics is needed. There are only a few indicators, and each answers a question a decision gets made on: is the program working, what does it cost the company and what to do about promotions next.
How many customers are enrolled in total and how many joined during the period. The base figure everything else is calculated from — and the first one that stops growing if enrolment stops.
How many members made a purchase during the month or quarter. The gap between this figure and the total database shows the program's real size: a database of ten thousand with three hundred active members is three hundred customers.
What share of members came back and how many times. The key indicator of a loyalty program: if it does not change, everything else is just handing out discounts in different packaging.
How many points were issued during the period — this is the company's liability in money. The figure that shows the cost of the program before customers come to redeem.
How many points were actually spent and what share of those earned. Low redemption is not a saving but a sign that customers are not coming back or do not know about their balance.
Which mechanics produced participation and revenue and which went unnoticed. The meaningful comparison is not between promotions but against an ordinary period without them.
How much a program member spends and how often they come — separately from those who buy without a card. That is the direct answer to whether the program pays for itself.
Reports export to a file, can be built on a schedule or pulled by an external system through the API — for when the company's consolidated reporting is kept in another program.
The bars show the share of the database, not of revenue. The line to examine in a table like this is the last one: of the three and a half thousand who bought, a third used their points — which means the rest either do not know about their balance or do not understand how to spend it. That is a job for notifications and the customer interface, not a reason to raise the earning rate.
This is the key section of the page. A loyalty program must not exist separately from purchases: if points are calculated in one place, receipts are rung up in another and the website's orders live in a third, the program turns into one more spreadsheet somebody fills in by hand. It only works as part of the business's overall digital system.
The first four steps happen before the receipt is closed and take fractions of a second — otherwise the queue at the checkout will feel the integration before the customer feels the program. The fifth and sixth run afterwards and do not affect the speed of the calculation. If the loyalty program is unavailable, the checkout must close the receipt without points rather than stop: the sale matters more than the accrual.
An order on the website can earn points just as a purchase in the shop does, and the balance can be shown in the buyer's account. How the storefront itself and order tracking are built is covered on the e-commerce software page.
It can work together with the retail layer: the receipt brings the contents of the purchase and the point of sale, and the loyalty program returns the accrual and the permitted redemption. In detail — on the retail software page.
The key connection: without it identification and redemption at the checkout are impossible. The scope of the exchange depends on what the specific checkout can do — this is established during discovery, before work begins rather than after.
An integration with the client directory is possible: the member card and purchase history are available to the manager, and enquiries and deals are visible next to the points account. The exchange runs on the client identifier.
It can work together with the company's accounting layer: the product list and categories come from there, and the points earned are reflected as a liability. The direction of exchange is determined by which set of records is recognized as the primary one.
The company's app can receive the balance, tier, offers and the code for the checkout through the same interface as every other system. No separate database is created for the app.
Messages to the customer, mailings, payment services, gift certificates. Every connection is a separate exchange module rather than a checkbox in the settings; we do not claim ready connections to specific services in advance.
The system's own interface: find a customer, request a balance, credit and redeem points, reverse an operation, pull the history. Everything without a dedicated module connects through it.
The exchange rules are the same everywhere: every operation has a key, so resending a receipt does not credit points twice; cancelling a purchase reverses both the accrual and the redemption; every message and every response is written to the exchange log. Without those three rules the customer's balance diverges from the receipts in the very first month, and trust in the program is lost faster than it is built.
A loyalty program makes sense where a customer can come a second time. If a purchase is one-off by nature, points will not work — no amount of configuration will change that.
Groceries, cosmetics, household goods, pet supplies — anywhere a buyer comes regularly. The most direct scenario: a short purchase cycle and a clear benefit from what has been accumulated.
The buyer is already signed in, so identification happens without a separate step and the balance is built into checkout. Here points compete with leaving for a rival shop in another tab.
Several locations and a shared customer database: a person accumulates at one shop and spends at another. This is also where the answer appears to which location brings in new customers and which only serves them.
A car service, a salon, a clinic, a workshop — anywhere visits repeat at a known interval. Often what matters is not the points but the history of visits and a reminder about the next one.
High visit frequency and a small receipt. The mechanic is usually simple — accumulating towards the next order or an offer during certain hours; complex tiers do not read here.
Wholesale supply, subscriptions, regular orders. The program is built not around points per receipt but around purchasing history and terms that depend on volume.
A loyalty program is not launched in full in a single day: while the rules have not been tested on real purchases, they will have to be changed in front of customers, and that is the worst way to do it. So go-live proceeds in stages, and each one builds on a working predecessor.
Who your buyers are and how often they come back, what is known about them today, which checkout is installed at your locations, whether there is an online store and an app. The outcome is a description of the process and the boundaries of the program, not a list of wishes.
The identification method, the earning rate, the redemption cap, the life of points. The most underrated stage: rules invented after launch have to be explained to customers who have already got used to different ones.
One shop or one shift goes through the full cycle on real purchases: enrolling a customer, identification, earning, redemption, a return, resolving a dispute. This also shows how much time has been added to service at the checkout.
The remaining locations follow the proven pattern, then the online store, roles and rights, then integrations as separate exchange modules. From there the history accumulates, and the data for tiers, segments and personal offers appears.
Tell us how many points of sale you have and how often customers come back, what is known about your buyers today, which checkout is installed and whether you have an online store. We will tell you which loyalty mechanic makes sense here, what could be launched first and what it would cost.
