Skip to content

Real-time engine for complex promotions, decided at payment

Qurable’s promotion engine decides complex promotions in real time, in ~150 ms (stackable, buy X get Y, bundles, tiered, basket, and bank or card conditions), at the POS, online, in the app, and at the kiosk. Your team sets the goal, the limits, and the budget; Qurable decides, applies the promotion at payment, and records the trace and cost of every decision.

See the difference

Fuel, restaurant, and retail programs already decide their rewards at payment with Qurable.

Read the MostaClub case
  • ~150 msReward decisionInside the authorization
  • < 50 msMember identificationp95, server-side
  • 99.5%SLACommitted availability
  • 3,000/sPeak in a single programMostaClub: 2.7M requests in 15 min

The difference

One promotion for everyone or the right promotion for each customer.

Choosing who gets a promotion is half the job. The other half is applying it at payment, protecting the budget, and knowing what it cost.

A promotion set up the same for everyone versus a promotion decided per customer at payment
Same promotion for everyoneDecided at payment, with Qurable
Who gets itEveryone who goes through the registerThe right customer, at the register (QR, wallet pass, or card), in your ecommerce, or in Qurable's white-label checkout
When they get itAt the register, the same discount for everyoneAt payment, decided in ~150 ms, or afterward, as a coupon for their next purchase, based on their profile
Who runs itOne tool picks who, and another system applies the couponThe same engine decides, applies it at payment, and deducts it from the budget
If several applyThey stack, or the first match winsThe best one available, with priorities and exclusion groups
Complex promotionsOne simple discount at a time, or custom development for each combinationStackable, buy X get Y, bundles, tiered, basket, and bank or card conditions, evaluated in real time
Channels and integrationTied to the POS or to a vendor portalAPI-first: the app, ecommerce, POS, and kiosk use the same engine through the API
BudgetKnown at month-end closeCaps per campaign, checked before the benefit is granted
How it's measuredBy redemptions, and tested in productionSimulated before launch and measured against a control group

From data to results

From each customer's data to the promotion that pays off.

Qurable decides at every purchase which promotion to give, to whom, and when, and shows you the incremental sales and what it cost.

  • 01 · Data + AI

    Behavioral audiences

    What they bought, in which channel, with which payment method, and their tier; AI builds one-to-one audiences.

  • 02 · AI Journey

    The best promotion, at payment

    With the receipt and the profile, AI Journey picks the promotion in ~150 ms, within your priorities.

  • 03 · Simulation

    The cost, before you spend

    Simulated on the last 7 days of real data: how often it fires and what it grants.

  • 04 · AI Journey

    AI builds it, your team approves

    Describe the campaign in one sentence and it's built; nothing goes live without your approval.

How it decides

Which promotion each purchase gets, and when to give it.

At every purchase, the engine filters the campaigns that qualify, applies your priorities and budget, and, based on the customer's profile, decides whether the reward comes at payment or on the next purchase.

CamilaRegister purchase · $45 · QRRecorded
  • 2-for-1 on the coffee brandCo-funded campaign · priority 1Inside the cap
  • 15% off her favorite categoryMoves to her next purchase, to bring her backNext purchase
  • 10% off the whole purchaseSame exclusion group as the 2-for-1, lower priority
  • Birthday couponDoesn't qualify: not her birthday month
  • Weekend promotionDoesn't qualify: the purchase is on a Thursday

Cost and the brand's contribution in the ledger, with the trace of which condition matched and what didn't fire.

Simulation based on a hypothetical case.

Camila’s case

Camila does her grocery run. She gets the promotion that fits her.

A supermarket, a coffee brand that wants new buyers, and a campaign co-funded by both. The promotion doesn't go to everyone: it goes to people who buy coffee but not that brand yet.

Camila: shows her QR at the register and pays $45

The register doesn't change: the POS sends the receipt via API.

Qurable receives the itemized receipt and the payment method

CCamila$45 · QR0 msRegister · supermarketReceipt$45IdentificationMember QR
01 / 04

Simulation based on a hypothetical case.

Promotion types

For everyone, for members, or one-to-one, in the same engine.

Every promotion has its own audience, budget, and cap. All four live in the same program and are measured the same way.

  • For everyone

    A general promotion at the register or online, with its budget and cap.

    Example
    10% off the whole purchase on Thursdays.
  • For members

    A member price for whoever identifies at the register, in the app, or online.

    Example
    Club members: special price on coffee.
  • Targeted, one-to-one

    For a behavioral audience, at payment or as a coupon for the next purchase.

    Example
    15% off her favorite category.
  • Co-funded with a brand

    A supplier brand pays part of the promotion and its contribution is recorded per campaign.

    Example
    Brand 50% and retailer 50%.

How it works

What it solves for each team.

Marketing wants to launch fast, finance wants to know the cost, and the register can't slow down. The same engine answers all three.

  • Audiences by purchase

    Built from what each customer bought and the attributes your program defines, for example people who buy the category but not the brand.

  • Coupons by day and time

    Coupon books with validity dates, weekdays, and time windows, plus a cap and budget per campaign.

  • AI Journey

    Describe the promotion in one sentence and it drafts the rule, audience, and coupon book from your catalog; your team approves it.

  • Referrals and contests

    Referrals that reward both sides, and contests with their own rules, on the same engine.

  • Capped budget

    Budget per campaign, limits per journey, and exclusion groups, checked before the reward is given.

  • Test before scaling

    Simulation on the last 7 days of real data, A/B tests, and a control group that doesn't get the promotion: you see the extra sales, not just the redemptions.

  • Each brand's share

    When a supplier brand funds part of the promotion, its contribution is recorded per campaign.

  • A trace of every decision

    Which event arrived, which condition matched, which effect applied, and what didn't fire and why.

  • No POS change

    API integration with the POS and ecommerce: Qurable decides the promotion and the register applies it.

  • Single-use coupon

    Single-use QR coupon validated at the POS.

  • The sale never stops

    If the service doesn't respond, the sale still goes through.

  • Every event, by webhook

    Coupons redeemed, points earned, and missions completed go out as signed webhooks to your systems.

  • No stack rebuild

    Connects to your POS, ecommerce, CRM, and messaging tool; it replaces none of them.

  • Your data, in your warehouse

    Every promotion, redemption, and contribution is exported daily to your data warehouse.

The pieces

How a promotion is built.

From audience to measurement, every piece lives in the same engine as the points program, cashback, and games.

PieceHow it works
WhoWhat each customer bought, their tier in the program, and the attributes your program defines.
IdentificationA QR that changes every 30 seconds, an Apple Wallet or Google Wallet pass, a card, or an account, in under 50 ms (p95).
ConditionsThe itemized receipt, payment method, channel, validity, weekday, and time window.
Across campaignsPriorities and exclusion groups: the best available promotion applies.
EffectsDiscount, coupon, points, partner points, badge, or an email or WhatsApp message.
At payment or afterA discount at payment or a coupon for the next purchase: Qurable picks the moment based on the customer's profile, aiming to bring them back.
After paymentA reward for the purchase or an upsell on the purchase confirmation screen.
ControlBudgets and caps per campaign, limits per journey, and program guardrails.
Who paysThe retailer, or the retailer plus one or more brands, with each one's share per campaign.
SimulationOn the last 7 days of real data: how often it fires, how often it doesn't match, and what it grants.
A/BTwo versions of the same promotion head to head, in the builder.
ApprovalNothing that affects members goes live without human approval.
TraceEach decision stores which event arrived, which condition matched, and which effect applied.
DataSigned webhooks and a daily export to your data warehouse.
IncrementalityThe control group is designed with each program, starting in the pilot.

More than discounts

One engine for every way to reward.

Rules, missions, games, and journeys use the same effects, budget, and ledger: adding a mechanic isn't a new project.

  • Gamification

    Spin-the-wheel, scratch cards, and five more mechanics, plus missions, badges, and stamps, with the prize decided before the animation.

    Timing
    A game reaches your app in less than a week.
  • Supplier-funded promotions

    The brands that sell to you pay part of the promotion and see their contribution per campaign.

    Example
    Brand 50% and retailer 50%.
  • Shared coupon books

    Share a coupon book with a partner brand, with a term sheet stating which part of the cost each one pays.

    In practice
    The partner accepts or rejects it from its console.
  • Cashback

    When the customer pays by card, the reward can be cashback decided inside the authorization.

    Live
    With Pomelo, Santander Peru, and Banisi.
  • Points and tiers

    Bonus points, multipliers, and missions on the points program, with every movement on the ledger.

    In practice
    Double points for a segment, with a cap.
  • Measure the result

    Simulation, A/B tests, and a control group to show the incremental sales from each promotion and what it cost.

    Guide
    Incrementality, cost, and ROI.

How we start

You set the goal, Qurable runs it and measures it.

We start with one real campaign before deciding the full scope. Then, in stages: more campaigns, supplier brands, and games. Live in 1 to 3 months.

Questions

What people ask about the promotion engine.

How it decides, how it's controlled, and how it connects.

It's the software that decides which promotion applies to each purchase: a discount, a coupon, bonus points, or a partner reward, based on rules, budget, and the customer. In Qurable that decision is made at payment, in ~150 ms, and each decision leaves a trace of which condition matched and which effect applied.

Yes. The engine evaluates, in ~150 ms, stackable promotions, buy X get Y (2-for-1, 3-for-2), bundles, tiered discounts, basket rules, and payment-method conditions such as the bank or card. MostaClub applies them in its app checkout, at the POS, and at the self-service kiosk, and Shell Box uses them with bank and card conditions. At peak, MostaClub alone processed 2.7 million requests in 15 minutes, about 3,000 per second.

Yes. The whole engine is used through the API: the app, ecommerce, POS, and kiosk ask the same engine for the promotion and get the decision in real time. The API is public and accessed with authorized credentials, and every promotion applied goes out by signed webhook.

One that decides at payment with the customer identified, checks budgets and caps before giving the reward, integrates with the POS without replacing it, and records each supplier brand's contribution. Qurable meets all four and runs fuel, restaurant, and retail programs in Latin America.

A points program builds value for future purchases; a promotion engine decides the reward for this purchase. In Qurable both share the same customer profile, rules, and ledger, so a promotion can give bonus points and a game can reward with a coupon.

Qurable picks the best available promotion based on each campaign's priorities and exclusion groups. The decision is stored: which event arrived, which condition matched, and which effect applied, so if a customer complains, you know why one applied and not another.

Updated