The problem, and whose problem it is

Unspent points are a liability on someone's balance sheet and a dead experience for their member.

• The client's problem. The points sit unspent. The liability stays on the books and the member never comes back.
• Why they will not fix it themselves. Commerce is heavy. Sourcing, warehousing, shipping, returns, and reconciling a split payment.
• The member's problem. A balance they can see and cannot use on anything they actually want.
• The supplier's problem. No route to a high-intent buyer who is already holding budget.
• What the market offers today. A commerce platform that cannot read a loyalty ledger, or a loyalty engine with nothing to sell.

Why Merit, and why now

The wedge is the end-to-end combination, anchored in MENA logistics. No single piece of it is the moat.

• It is already running. Al Fursan is live on Merit today: physical goods at scale, gift cards, the Apple Store, and an earn model.
• The commercial shape. 16 large tenants and 100 sellers by December 2026, roughly $869.2K per month recurring at OKR close.
• Why the rebuild, not a patch. The Old World was built business-led with no product team steering it, so every tenant is an engineering project.
• What changes that. Storefront Builder. It turns tenant number two into a configuration exercise instead of a build.
• The honest risk. No client runs on the New World yet, and the Old World keeps moving while the rebuild catches up.

The gap Merit sits in

Airlines, banks and telcos hold points their members cannot spend. Running commerce is heavy. Merit runs it for them.

Three roles, and one word that means two things

The vocabulary is genuinely confusing on day one, because internal teams use the same word differently.

• Client, or tenant. Owns the loyalty points. Al Fursan, SAIB, STC, Entertainer, Nielsen. Wants members to burn.
• Merit. Owns the commerce stack and the supplier network. Runs catalogue, orders, logistics, payments and settlement.
• Supplier, or seller. Owns the goods. Almania, Aleph, Just Lounge, Amazon, STC as an Apple distributor.
• The end customer. The client's member. They never see Merit branding. They think they are shopping with Saudia.
• The landmine. Platform engineering calls everyone on the system a tenant, sellers included. Check which side someone means.

Five earning layers, and two of them are not software

Two layers behave like software and three behave like retail. They must never be blended in a P&L.

• SaaS platform fee. Recurring per tenant, indicatively $500 to $20,000+ per month by tier. Target around 80% GM.
• GMV commission. Take rate on transaction value, roughly 1% to 3%, descending by tier. Target around 80% GM.
• Goods margin. Wholesale against retail spread where Merit is Merchant of Record. Retail economics, 8% to 20%.
• Interchange and FX. Thin margin on the cash leg of a mixed payment, and on cross-currency settlement.
• Ad placements. Suppliers pay for featured slots. A dynamic slot auction is planned for 2027.
• Open, with Finance and Fred. Do we book full GMV as revenue under Merchant of Record, or only the commission we keep?

Read the map by what each service knows

A set of decoupled services behind a white-label storefront, with one integration boundary in the middle.

What is built, and what is still a promise

Status as of July 2026. The honest version, not the roadmap version.

• Live. PIM. Pricing and Commission Engine, done. Seller Portal, live and expanding.
• Close. OMS is in final integration. Storefront Builder is in review. Storefront sits at roughly 50 to 60 percent.
• Building. E-Commerce Core. Search, which Fred has ruled launch-critical rather than optional.
• Not started. TMS build starts Q3. Seller payout and settlement in Q3 2026. Order fraud screening is still to do.
• Partial. Notification Hub. SendGrid is the live connector and SMS is still being wired.

Two platforms, and both of them are moving

The single most confusing thing about Merit for a newcomer, Confluence included.

One product, many offers

The core data-model idea. Everything about the Buy Box, catalogue quality and supplier onboarding follows from it.

Why seller onboarding is the hard part

A supplier cannot simply create a product and set a price. The catalogue would fill with duplicates.

• Exact match first. Regex and exact string matching. Cheap, fast, correct when it fires. This is what is live today.
• AI match second. A model judges whether the upload is the same thing. Approved, not confident, or different product.
• Human review on the middle case. The not-confident bucket is where ops time goes, and where a better model pays for itself.
• Bulk upload, confirmed 21 Aug 2026. Row isolation already exists. 500 failed rows out of 5,000 do not drop the other 4,500.
• The queue is already live. Valid rows publish only on an exact product and variant match. The rest go to review.

The Buy Box, in order

When several suppliers offer the same variant, something has to pick one. The rule was locked on 28 July 2026.

An order, end to end

From browsing to points landing back in a member account, using the Al Fursan retail flow as the reference case.

The part Shopify cannot replicate

Mixed payment is Merit's proprietary piece, and it is the reason clients sign.

• Points only. The whole basket is paid in points, burned through Point Exchange, with no cash leg.
• Mixed payment. Points and card in one transaction, split on a slider. It removes the wall on high-value items.
• Earn on purchase. Calculated from what was actually paid, held pending through the return window, then credited.
• Rate lock at fulfilment. The rate applied is the rate at fulfilment, not at browse time. It protects the page price.
• Merit validates, because the client often will not. Some loyalty APIs mint points on request, with no double-entry ledger.
• A fourth pattern with no points at all. Bank offer redemption, as in SAIB, is entitlement-based and capped per customer per month.

The names you will hear in every standup

One company can appear on both sides of the business at once. STC does.

• Al Fursan. Saudia's loyalty program. The flagship and the most complex. Physical goods, gift cards, Apple Store, earn model.
• Entertainer, SAIB, Nielsen, Kantar. An embedded marketplace, entitlement-based offer redemption, and gift cards as survey incentives.
• STC. A client for Qitaf points, and an Apple distributor supplying the Al Fursan Apple Store.
• Almania and Just Lounge. API connected, and both are high out-of-stock contributors because the stock data is unreliable.
• Salasa and OTO. Salasa is modelled as a self-shipping supplier. OTO is logistics aggregation, not goods.
• The long tail. Roughly eight manual suppliers, decrement status still being confirmed.

Four openings, ordered by how ready they are

Product matching is ready now. Everything else needs data or scrutiny first.

• Ready now, product matching. Exact match is live and the model layer is deliberately unfinished. Clear ops cost attached.
• Needs scrutiny, the existing fraud model. Precision, recall and accuracy all near 100%, which reads as overfitting. Test set undocumented.
• In flight, Arabic-first search. Typo-tolerant query understanding on ElasticSearch plus AI. Launch-critical.
• Later, recommendations and ranking. Blocked on behavioural data volume, so build starts in 2027.
• Two platform bets. API parity for every capability, and an MCP companion so a supplier can integrate conversationally.

What to hold in your head

Six sentences that carry most of the product.

• PIM knows what a product is. E-Commerce Core knows who is selling it and at what price.
• OMS knows what happened to the order. TMS knows where the box is.
• A product is shared. An offer belongs to one seller. The Buy Box picks between offers.
• Trusted stock is the first gate, because an untrusted stock number is a future cancellation.
• Old World runs every paying client. New World has none yet, and both keep moving.
• Mixed payment is the thing no competitor has, and it is why clients sign.

The problem, and whose problem it is

The value is real, it belongs to the consumer, and it is unreachable.

• The consumer's problem. Airline miles in one app, bank points in another, telco points in a third. None of them talk.
• The size of it. In KSA alone an estimated SAR 2.4B of points expire unused every year.
• The issuer's problem. Idle point liability on the balance sheet, and a program that paid to acquire someone who never came back.
• The merchant's problem. No clean route to a buyer who is high-intent and already funded.
• Why nobody has closed it. Closing it needs identity, a ledger, an exchange and a checkout. Each competitor holds one piece.

Why Merit, and why now

B2B sales cycles are long and the client always owns the customer. Merit needs a moat it controls.

• It is shipped, not a deck. Live on both stores. Android since March 2026, iOS since 30 July 2026. Gift cards, marketplace and payments operational.
• It reuses what already exists. Same supplier network and same commerce core as the B2B side. The consumer app is the new part, not the machinery.
• The commercial shape. Around $50M GMV in year one across four markets, with a December 2026 checkpoint of 125K monthly transacting users.
• Where the real margin is. mTrust sold as Data-as-a-Service is the highest-gross-margin line in the whole model.
• The honest risk. Organic execution alone caps the business at $3M to $4M ARR, and full any-to-any exchange is a roadmap claim today.

Many balances in, one balance out

The E-Commerce Solution sells infrastructure to loyalty programs. This one sells to the person holding the points.

Four roles, and the consumer is now ours

The important shift from the B2B side is that the consumer is Merit's own user, not somebody else's member.

• Issuer. Banks, airlines, telcos, retailers. They hold idle point liability and earn a share when their points get spent.
• Merit. Owns the wallet, the identity, the exchange and the checkout.
• Consumer. Points-rich, digitally active, in KSA or UAE. Sees Merit branding and holds a relationship with Merit.
• Merchant. Gift-card brands, marketplace sellers, and later offline merchants accepting QR payment.
• The landmine. An issuer here is often a client on the other side. STC is both, and an Apple distributor as well.

Six streams, and the revenue definition trips everyone up

Revenue is full transaction value, GMV. The take rate is a margin metric, not the revenue metric.

• 3P transaction take rate. Percentage of points spent or traded, blended around 8%. Software-like economics.
• 1P merchant margin. 25% to 30% on first-party goods, dark kitchens and exclusives. Real retail. Model it separately.
• Issuer revenue share. A pass-through split. Net take is after the issuer's share.
• Data-as-a-Service, mTrust. API access sold to financial institutions. Very high gross margin.
• Point top-up spread and Merit Elite. Spread on points bought or converted, plus a premium membership at around SAR 299 per month.
• If someone quotes 8% revenue. They are quoting margin. Correct it, because the two numbers are not comparable.

What is live, what is building, what is a promise

Status as of early August 2026.

• Live. The mobile app on both stores. Wallet and Ledger. Gift cards and digital goods. The physical-goods marketplace.
• Shipping now. Mixed payment v1, on the reserve, charge, commit saga.
• The critical Q3 dependency. The LMS loyalty stack. Integration in August, then cashback, referral, gamification and P2P.
• Highest ship risk. QR Scan-to-Pay. It needs Merchant Locator, POS integration and a banking ledger.
• Later. mTrust GA in October. Point Exchange beta in March 2027. Virtual card in January 2027. Token production in 2027.
• Hard, not nice-to-have. Mixpanel. Every OKR is unmeasurable without it.

The one architecture rule

If you internalise a single thing about how this product is built, make it this one.

Three verbs, at three very different stages

Burn works today. Earn is the Q3 build. Exchange is the differentiator and the least built.

• Say this plainly. Any claim that Merit does full any-to-any exchange today is a roadmap claim, not a product claim.
• Read-only aggregators. AwardWallet can see your points and cannot spend them.
• Single-program embeds. PointsPay can spend one program's points and aggregates nothing.
• Card optimisers. MaxRewards routes a card choice and touches no loyalty balance. Curve consolidates cards, not points.
• Regional enterprise vendors. Dsquares and TADA sell into clients, so their programs stay siloed per client by design.
• The whitespace. Aggregate and spend, in one place. Nobody in the field covers both columns.

Three rails at checkout

The checkout is where the product either delivers on its promise or does not.

• Points only. The whole basket from the aggregated balance. LMS burns, no cash leg. Login required, because points are account-bound.
• Mixed payment. Points and card in one transaction. Merit's proprietary piece. v1 is a split, and the slider is v2.
• Cash only. Card, Apple Pay or Google Pay through Moyasar. The only rail that can work without an account.
• So guest checkout is fiat only. A guest buys with a card, the order keys to a contact, and the account offer comes after.
• The conversion argument. Every gate before the money is a gate some people will not walk through.

Reserve, then charge, then commit

Two ledgers, one transaction. If either leg fails, the other has to unwind.

Nine stages, one gift-card purchase

Everything here is inside the app and the Superplatform services it calls.

• 1 to 3. Browse as a guest. Sign in with Merit ID. The wallet resolves one number out of several connected programs.
• 4 and 5. Choose the rail. Reserve the points, charge the card through Moyasar, then commit the burn.
• 6 and 7. Fraud screening and the mTrust signal apply. A digital good is issued immediately.
• A physical product diverges here. It hands off to the commerce stack, which is a separate product line with its own team.
• 8 and 9. Notify through the shared service. Cashback on the first three orders, 5%, then 8%, then 12%, credited by LMS.
• The app fires the trigger. LMS issues the points. That split holds for streaks, spins and badges too.

Year one is KSA first, then UAE, then Jordan

Each market is a different payment rail, a different regulator and a different point-issuer landscape.

A launch is not a translation

Nine workstreams open for a new market, and only one of them is language.

• Three can block a date outright. The payment rail, registration and OTP delivery, and the regulatory position.
• The rest is work, not a gate. Issuer contracts, catalogue, support, marketing, pricing, and yes, language.
• KSA, on the token. Blockchain production sits outside virtual asset regulation. The token is internal and closed loop only.
• What holds that position. Transfer-restricted, spendable only inside the Merit merchant network, with no fiat conversion in the product.
• Jordan, open with counsel. Loyalty overlay, or an EPSP licence? Data residency and KYC apply per market.

Organic alone caps the business

The gap to the target is closed by three stacking paths, and each one is gated on a decision rather than on engineering.

Two named products, and two openings

mTrust does double duty: it protects the transaction, and it sells as an API.

• mTrust Score. A 1 to 100 behavioural trust signal. Engine GA targeted October 2026, owned by Data Platform.
• B2C owns the display only. Badges, streaks, leaderboards and score-based perks. Not the model.
• Why it matters commercially. Sold to financial institutions, it is the highest-gross-margin line in the whole revenue model.
• AI Purchase Advisor. Which card, or points, maximises value on this purchase. MaxRewards already has a comparable, so it is parity.
• Personalised offers. Blocked on behavioural data volume. This follows instrumentation rather than leading it.

What to hold in your head

Six sentences that carry most of the product.

• Aggregate points into one wallet, then make that wallet spendable. That is the whole product.
• The app owns the surface. LMS owns the points. There is no third layer underneath.
• Reserve, then charge, then commit. Never burn before the card clears.
• Revenue is GMV. The take rate is margin. They are not the same number.
• Burn works, earn is the Q3 build, and exchange is a roadmap claim today.
• Every growth path is a decision for Fred, not an engineering task for the team.

Julie holds six programs and cannot say what they are worth

Identity does not travel, so value cannot travel either. She runs product for a loyalty company, and she still cannot use what she holds.

SAR 400 at checkout, and Merit finds SAR 91 the consumer forgot

This is the part a consumer feels. Everything else is infrastructure under it.

Merit ID is the spine. The Super App is one thing standing on it.

Identity, then connections, then the value graph, then orchestration, then settlement.

Every arrow in is a partner Merit never has to verify twice

Density is the asset, and the connection primitive is already built and working.

The starting position, stated plainly

A vision that overstates the present tense stops being useful the first time somebody checks.

• Live. The identity system of record, off-chain. Auth, KYC tiers, sessions and security, all built.
• Already connected, on the legacy platform. Dozens of programs. That estate is real, and the new layer inherits it. Merit is not starting from zero connections.
• Live but untested. The Merit ID app is in the KSA and UAE stores. Deliberately not promoted, still in final testing.
• Built. The issuer connection journey end to end, and bulk identity creation, proven on the Gift Global migration with 1,341 users.
• One. Issuers visible in the app: Peoppl. BSF, Al Rajhi and ANB are connected, and a consumer cannot see any of them yet.
• In build or draft. mTrust, GA October 2026. Membership BRD at v0.1. The checkout agent is a concept with no build sized.

Now, connect, orchestrate, intelligence

Twelve months, four stages, and each stage is useful on its own.

Every new connection makes every existing one worth more

Seat revenue scales with customers. Orchestration economics scale with the connections between them.

• The monetisation layers. Identity and API fees, partner integration, transaction and orchestration economics, Point Exchange economics, mTrust, and premium capabilities.
• The one that compounds. Cross-issuer orchestration and settlement. It improves with network density rather than with seat count.
• The moat. The tenth client is cheaper than the first, and each new partner makes the network denser for everyone already on it.
• The wedge. Jahez, because no merchant refuses Jahez. The unlock is entering under a national initiative such as NCNP or Nusuk.
• The pressure. Sanabil is building in the same space, and contacts across Visa, Google Pay and Apple Pay call this the next thing they want to see.
• What is still open. Willingness to pay is not validated. Price points are in the appendix and are not a commitment.

Indicative pricing, not yet validated

Shown to size the opportunity. Not a commitment, and not for external use.

Potential on-chain architecture, subject to legal validation

Blockchain has to solve a problem here. It is not part of what Merit ID is.

Embedded finance is optionality, not a capability

Interesting future use cases. None of them is an established Merit ID capability today.

• Lending against aggregated value. If Merit ID can see what a member holds, a lender could in principle advance against it. Untested, unbuilt and regulated.
• Tokenising loyalty liabilities. Points sit on an issuer's balance sheet. Liquidity against them is strategic optionality, not a proposition to sell.
• Why SAIB is interested. They want equity rather than a supplier contract. The interest rests on the aggregation, and the aggregation is what gets built first.
• mTrust. The 1 to 100 score, sold to issuers as a risk-signal API. GA targeted October 2026. This one is real and it is on the roadmap.
• Membership. Shaped for Riyadah as a microsite, then made a native Merit ID feature, because membership is status and entitlements on an identity Merit already owns.
• Why it stays out of the main deck. Merit does not have to become a financial institution to have an enormous addressable opportunity.

Who is who, and the words that trip people up

The government identity source separates this from an auth product. The issuer separates it from a loyalty engine.

• Issuer. Banks, telcos, airlines, marketplaces. They connect once and stop re-verifying.
• Identity provider. Nafath, for Saudi national identity. It feeds high-assurance verification, not points.
• Merit ID. Owns the verified identity, the consent layer, the entitlement graph and the orchestration. It does not own the underlying value.
• Member and User. Only Members have a Merit ID. A User is a corporate admin on normal corporate auth, and that is a different thing.
• Program. The B2C app, a client marketplace, an LMS tenant, a membership. It issues status on an identity it did not build.
• Connected and visible. Connected means a technical link exists. Visible means a consumer can see it. Four issuers are connected and one is visible.

Permission, then token, then live balance

This journey is already built and working. It is the primitive the whole network is made of.