Payment Processing Basics

TL;DR: We built a transformer-based payments foundation model. It works. For years, Stripe has been using machine learning models trained on discrete features (BIN, zip, payment method, etc.) to improve our products for users. And these feature-by-feature efforts have worked well: +15% conversion, -30% fraud. But these models have limitations. We have to select (and therefore constrain) the features considered by the model. And each model requires task-specific training: for authorization, for fraud, for disputes, and so on. Given the learning power of generalized transformer architectures, we wondered whether an LLM-style approach could work here. It wasn’t obvious that it would—payments is like language in some ways (structural patterns similar to syntax and semantics, temporally sequential) and extremely unlike language in others (fewer distinct ‘tokens’, contextual sparsity, fewer organizing principles akin to grammatical rules). So we built a payments foundation model—a self-supervised network that learns dense, general-purpose vectors for every transaction, much like a language model embeds words. Trained on tens of billions of transactions, it distills each charge’s key signals into a single, versatile embedding. You can think of the result as a vast distribution of payments in a high-dimensional vector space. The location of each embedding captures rich data, including how different elements relate to each other. Payments that share similarities naturally cluster together: transactions from the same card issuer are positioned closer together, those from the same bank even closer, and those sharing the same email address are nearly identical. These rich embeddings make it significantly easier to spot nuanced, adversarial patterns of transactions; and to build more accurate classifiers based on both the features of an individual payment and its relationship to other payments in the sequence. Take card-testing. Over the past couple of years traditional ML approaches (engineering new features, labeling emerging attack patterns, rapidly retraining our models) have reduced card testing for users on Stripe by 80%. But the most sophisticated card testers hide novel attack patterns in the volumes of the largest companies, so they’re hard to spot with these methods. We built a classifier that ingests sequences of embeddings from the foundation model, and predicts if the traffic slice is under an attack. And it does this all in real time so we can block attacks before they hit businesses. This approach improved our detection rate for card-testing attacks on large users from 59% to 97% overnight. Perhaps even more fundamentally, it suggests that payments have semantic meaning. Just like words in a sentence, transactions possess complex sequential dependencies and latent feature interactions that simply can’t be captured by manual feature engineering. Turns out attention was all payments needed!

#Africa bleeds $5B a year not to #corruption or #mismanagement, but just to move money within its own borders. Example: A Kenyan business paying a Ugandan supplier. Instead of Nairobi → Kampala, money goes: Nairobi → USD conversion (1–2%). USD routed via New York/London ($20–50 fee). USD → Ugandan shillings (another 1–2%). By the time a $26,000 invoice is paid, $500–1,000 is gone. Whilst we may be denied visas, our money travels freely through New York. And it’s not just trade: Africa’s #diaspora sends $95B home each year, yet pays the world’s highest remittance costs. -We pay the highest cost for credit. -We pay the highest cost for payments. -We pay the highest cost to send our own money home. It’s not inefficiency. It’s design. The #GlobalFinancialSystem wasn’t built for us. The good news? Solutions exist. #PAPSS (Pan-African Payment and Settlement System) is already live linking 15 central banks, 150 commercial banks, and 14 payment switches, with the capacity to handle $300B in intra-African trade annually. Through PAPSS, that same Kenya–Uganda  transaction could  look very different: -One direct conversion from KES → UGX (0.2–0.5% spread). -Settlement netted via African central banks. -Funds received in hours, not days. Estimated cost: $60–150.  Potential savings: $500–950 on a single $26,000 payment. No detours. Value stays in Africa. The challenge isn’t invention. It’s implementation. One Africa. One market. One #payment system. AI image below*

Stripe 𝗯𝘂𝗶𝗹𝘁 𝘁𝗵𝗲 𝗳𝗶𝗿𝘀𝘁 𝘁𝗿𝗮𝗻𝘀𝗳𝗼𝗿𝗺𝗲𝗿 𝗺𝗼𝗱𝗲𝗹 𝘁𝗿𝗮𝗶𝗻𝗲𝗱 𝗼𝗻 𝗽𝗮𝘆𝗺𝗲𝗻𝘁 𝗱𝗮𝘁𝗮! Not for text and NOT for code, BUT for billions of payments. Think GPT, but instead of learning language, it learned the structure, behavior, and patterns behind every transaction: ⬇️ 𝗛𝗲𝗿𝗲 𝗶𝘀 𝘄𝗵𝗮𝘁 Stripe 𝗷𝘂𝘀𝘁 𝗱𝗶𝗱? For years, Stripe used traditional ML — separate models for fraud, disputes, and authorizations. Each one relied on handpicked features like BIN codes, ZIP codes, email addresses, and payment methods. That worked — but it was narrow, manually intensive, couldn’t scale and most importantly, it missed the bigger picture. So Stripe trained a transformer, just like GPT — but instead of learning language, it learned from billions of transactions. Each payment — from a coffee in Paris to a subscription in Tokyo — was turned into a dense vector: a numerical fingerprint capturing its behavior and context. 𝗧𝗵𝗲 𝗼𝘂𝘁𝗰𝗼𝗺𝗲? ➜ Transactions with similar behavior cluster naturally — by issuer, merchant, location, or risk ➜ Suspicious patterns emerge organically — without handcrafted rules ➜ Fraud becomes easier to detect — not because it was labeled, but because it’s "understood" This foundation model captures now the structure and relationships between transactions — in real time — the way GPT models understand the flow of words in a sentence. Stripe no longer needs a different model for every use case. They’ve built one that generalizes across many — and keeps learning. 𝗧𝗵𝗲 𝗿𝗲𝘀𝘂𝗹𝘁? They tested it on one of the hardest problems in the space: Card testing attacks that hide in legitimate traffic. ➜ Traditional ML: 59% detection ➜ Transformer-based model: 97% — overnight Visionary work by Stripe! BUT this approach has implications far beyond payments. Great example to see that foundation models aren’t limited to text. The next phase of AI will probably focus more on transformer architectures trained on high-value, underexplored data domains: transactions, supply chains, behavioral signals, scientific processes — even spreadsheets. 𝗜 𝗮𝗺 𝗽𝗿𝗲𝘁𝘁𝘆 𝘀𝘂𝗿𝗲 𝘄𝗲 𝘄𝗶𝗹𝗹 𝘀𝗲𝗲 𝗺𝘂𝗰𝗵 𝗺𝗼𝗿𝗲 𝗱𝗼𝗺𝗮𝗶𝗻-𝘀𝗽𝗲𝗰𝗶𝗳𝗶𝗰 𝗳𝗼𝘂𝗻𝗱𝗮𝘁𝗶𝗼𝗻 𝗺𝗼𝗱𝗲𝗹𝘀 — 𝗽𝘂𝗿𝗽𝗼𝘀𝗲-𝗯𝘂𝗶𝗹𝘁 𝘁𝗼 𝗼𝗽𝗲𝗿𝗮𝘁𝗲 𝗶𝗻𝘀𝗶𝗱𝗲 𝗰𝗼𝗺𝗽𝗹𝗲𝘅 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 𝗹𝗶𝗸𝗲 𝗳𝗶𝗻𝗮𝗻𝗰𝗲, 𝗵𝗲𝗮𝗹𝘁𝗵𝗰𝗮𝗿𝗲, 𝗮𝗻𝗱 𝗲𝗻𝗲𝗿𝗴𝘆. 𝗙𝗼𝗿 𝘆𝗲𝗮𝗿𝘀, 𝗺𝗮𝗰𝗵𝗶𝗻𝗲 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗵𝗮𝘀 𝗯𝗲𝗲𝗻 𝗳𝗼𝗰𝘂𝘀𝗲𝗱 𝗼𝗻 𝗹𝗮𝗯𝗲𝗹𝗶𝗻𝗴 𝗽𝗿𝗼𝗯𝗹𝗲𝗺𝘀. 𝗡𝗼𝘄, 𝘄𝗲'𝗿𝗲 𝗲𝗻𝘁𝗲𝗿𝗶𝗻𝗴 𝗮 𝗽𝗵𝗮𝘀𝗲 𝘄𝗵𝗲𝗿𝗲 𝗺𝗼𝗱𝗲𝗹𝘀 𝗯𝗲𝗴𝗶𝗻 𝘁𝗼 𝘂𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱 𝘁𝗵𝗲𝗺 — 𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗮𝗹𝗹𝘆, 𝗰𝗼𝗻𝘁𝗲𝘅𝘁𝘂𝗮𝗹𝗹𝘆, 𝗮𝗻𝗱 𝗮𝘁 𝘀𝗰𝗮𝗹𝗲. Full story in the comments. 𝗣.𝗦. 𝗜 𝗷𝘂𝘀𝘁 𝗹𝗮𝘂𝗻𝗰𝗵𝗲𝗱 𝗮 𝗳𝗿𝗲𝗲 𝗻𝗲𝘄𝘀𝗹𝗲𝘁𝘁𝗲𝗿 𝗼𝗻 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁𝘀 𝗮𝗻𝗱 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀 — 𝗮𝗹𝗿𝗲𝗮𝗱𝘆 𝗿𝗲𝗮𝗱 𝗯𝘆 𝟮𝟬,𝟬𝟬𝟬+. 𝗝𝗼𝗶𝗻 𝗵𝗲𝗿𝗲: https://lnkd.in/dbf74Y9E

More merchants than ever are utilising the services of more than one acquirer. Dual acquiring isn't anything new, but the rise of orchestration platforms and redefined dynamic routing products has made it easier than ever for merchants to leverage more than one acquirer relationship. This approach allows merchants to diversify their payment processing channels, increasing failover resilience, taking a stronger stance on good relationship management and potentially optimising costs, if deployed in the right way. _______________________________ ➡️ How Dual Acquiring Works 🔹 Merchants integrate with two (or more) acquiring banks or PSPs, either directly or through a payment gateway or orchestration platform. 🔹 Transactions can be routed dynamically based on various factors such as cost, card type, or transaction volume. 🔹 Some merchants use automated systems to select the most cost-effective or reliable acquirer for each transaction in real time. 🔹 A layer beyond this, On-Us transaction selection is a further highly efficient functionality that should be considered by merchants. In an On-Us transaction, the issuing bank and the acquiring bank are the same institution. This enables them to maximise their profits from the transaction and provide pricing at rates lower than standard interchange fees for the end merchant. With an alternate Off-Us transaction (when the issuing and acquiring banks are separate entities), the interchange fee is divided between both banks, which results in lower profits for each and higher pricing for the end merchant. ➡️ Merchant Considerations 🔹 Your setup requires the ability to support multiple acquirer integrations within the payment processing system you roll out. This can be a slightly more complex initial integration, but it will lead to considerable benefits in the long run. Orchestration platforms are also making lighter work of this for merchants. 🔹 Managing settlements and reconciliations across multiple acquirers can add operational complexity. This is why we are seeing more and more 'payments teams' being created within merchant organisations. 🔹 Ensuring compliance with regulations such as PCI-DSS when working with multiple acquirers can be slightly more cumbersome. _______________________________ At Nomupay we understand that as well as being an acquirer in our own right, maintaining a close working relationship with a whole variety of other acquirers globally is the right thing to do for our end merchants. That's why we also lean into a gateway agnostic approach. A merchant will always appreciate the best fit for them being recommended, even if that means relinquishing some share of wallet over the long term. More acquirers need to be thinking this way in 2025. So, if you are a payments provider interested in exploring a partnership with Nomupay, please do drop me a message. 📨

What Is Payment Orchestration? 💡 Payment orchestration is the process of streamlining and managing all payment flows through a centralized platform. It connects to multiple PSPs, acquirers, and payment methods, enabling businesses to scale, process transactions more efficiently, and increase approval rates. Orchestration acts as the layer above PSPs and acquirers, managing how payments are routed, retried, and reported across the entire stack. At its core, orchestration is about building modular, adaptable, and scalable payment infrastructure. What Payment Orchestration platforms do: 1️⃣ Acceptance: Enables merchants to extend their acceptance by providing payers with the preferred ways to pay, like cards, wallets, and local payment methods, all via a single integration. 2️⃣ Routing & Fallbacks: Automatically sends each transaction through the best provider using real-time logic (country, card type, amount, past performance, or real-time availability). If one provider is down or a transaction fails, the orchestrator reroutes through an alternative, minimizing declines. 3️⃣ Reconciliation: - Pulls all settlements, fees, chargebacks, and refunds into one dashboard. - Data is reconciled in real-time and on an hourly basis to prevent any drift between the provider and orchestrator. 4️⃣ Reporting & Analytics:  - Gives full visibility into payment performance: approval rates, declines, and chargebacks across all markets and providers. - All data is available in a single admin panel. Benefits of Payment Orchestration: ✅ Multi-PSP setup: Easily add or switch providers without touching the code. ✅ Higher approval rate: Smart routing + fallbacks = fewer false declines. ✅ Lower costs: Route to local acquirers; skip unnecessary FX fees; A/B test providers to find the best blend of cost and performance. ✅ Global-ready: Expand to new regions by simply plugging in the right local PSP. ✅ Better checkout UX: Supports payment methods customers expect; fewer errors and drop-offs.  ✅ Monitoring: Provides real-time visibility into performance, errors, and approval rates across all providers. ✅ Resilience: No single point of failure: if one provider goes down, payments don’t. ✅ No-code workflows: One integration, no custom logic, just config. ✅ Adaptive fraud routing: Triggers 3DS only when it makes sense based on issuer, region, or amount. Source: Solidgate - https://shorturl.at/jeBjK #Innovation #Fintech #Banking #OpenBanking #API #FinancialServices #Payments #PaymentOrchestration #Acceptance #Acquiring #Scheme #Settlement #Transaction #Fraud  

Online checkout has come a long way, but surprisingly I find that some of the biggest frictions remain quite basic. To put it bluntly, a checkout that doesn’t match how shoppers expect to buy is still one of the fastest ways to lose revenue. Small details add up... ... a page layout that feels cluttered ... that breaks on mobile ... forms that ask for too much ... missing payment methods or currencies ... error messages that confuse rather than guide ... accessibility that isn’t truly accessible And the list goes on and on... Shoppers are now conditioned by one-click flows, digital wallets and auto-fill. Asking shoppers to type card numbers or recall passwords is a big no-no. In a world where competition is one tab away, every point of friction widens the gap between intent and conversion. But equally in today's context, the harder part for merchant is probably not not technology, it is noise. The paytech landscape moves quickly and is full of overlapping terms, converging features and categories that blur into each other. It is easy to get caught up in the volume of updates and miss the few that actually matter. Not every new payment trend or product update matters. What usually cuts through the noise is looking at changes through the lens merchants already operate with: 👉Cost: reducing fees or operational overhead 👉Conversion: lifting approval rates or reducing drop-off 👉Compliance: strengthening alignment with requirements and reducing exposure 👉Customer experience: making the buying flow faster, safer, or more intuitive The point is that innovation only becomes valuable when it actually impacts the overall business operation. Today, modern payment stacks give merchants the tools to compete in a very different way: flexible, modular and built for adaptation. For many merchants, the real shift comes from reducing dependency. Legacy setups lock businesses into rigid structures and single points of failure. A more modern architecture increases control, strengthens continuity and makes adaptation faster than the competitive set. And across all of this, security sits at the core. Trust and loyalty only grow when payments feel safe. More alternative methods alone won’t solve that. A good payment strategy simplifies where it matters, modernises where it counts and builds for resilience, trust and long-term customer relationships. I was pleased to share some thoughts on the recently released article by WIRED x Mastercard on payment innovation. The article explores how trends like click-to-pay, passkeys and the rise of agentic commerce are reshaping what seamless checkout really means. https://lnkd.in/dN7_gUy7 Mastercard WIRED Paypr.work [ˈpeɪpəwəːk] -- 𝘐𝘧 𝘺𝘰𝘶’𝘳𝘦 𝘭𝘰𝘰𝘬𝘪𝘯𝘨 𝘵𝘰 𝘣𝘶𝘪𝘭𝘥 𝘢 𝘥𝘦𝘦𝘱𝘦𝘳 𝘶𝘯𝘥𝘦𝘳𝘴𝘵𝘢𝘯𝘥𝘪𝘯𝘨 𝘰𝘧 𝘩𝘰𝘸 𝘱𝘢𝘺𝘮𝘦𝘯𝘵𝘴 𝘳𝘦𝘢𝘭𝘭𝘺 𝘸𝘰𝘳𝘬 𝘢𝘯𝘥 𝘢𝘱𝘱𝘭𝘺 𝘵𝘩𝘢𝘵 𝘬𝘯𝘰𝘸𝘭𝘦𝘥𝘨𝘦 𝘴𝘵𝘳𝘢𝘵𝘦𝘨𝘪𝘤𝘢𝘭𝘭𝘺, 𝘭𝘦𝘵’𝘴 𝘵𝘢𝘭𝘬: 👉 [email protected] 👉 www.paypr.work

Forget one size fits all, local payment methods are the new consumer favorites in markets round the world Local Eats Global, as global card schemes lose ground to local favorites like digital wallets A2A, carrier billing and BNPL in Ecommerce These are some findings from a The 2024 Global Ecommerce Report which analyses data from 37 major markets, highlighting global, regional and country specific trends. Key Findings ▸ Local payment methods will reach 58% of all ecommerce transaction value globally by 2028, reflecting a major shift within the ecommerce payments market. ▸ By 2028, almost 37% of all individuals globally will actively use local payment methods, reflecting massive growth and expansion of the ecommerce market across the world. ▸ Card values will decline to 20% of transaction value by 2028, from 31% in 2023, reflecting a major shift as the ecommerce market expands. ▸ BNPL is steadily growing its share of ecommerce values, from 4% in 2023 to 5% in 2028, reflecting steady progress outside of key, already highly saturated markets, such as Australia, Germany and Sweden. ▸ A2A payments are seeing strong growth, from 8% of ecommerce spend in 2023 to 16% in 2028, a dramatic increase, reflecting major shifts in this market. Why Going Global Needs Local Payment Solutions As someone who knows payments, I'll explain why understanding local preferences matters for success worldwide. Thinking Globally, Acting Locally: ▸ One-size-fits-all no longer works: Countries have very different ways to pay. If you ignore this, you'll lose sales. ▸ Welcome local favorites: Give your customers the payment methods they like. This shows respect for their choices and helps build trust. ▸ Give a variety of payment options: People enjoy choices when shopping online! In some places, folks use several ways to pay. Don't stick to just one. ▸ Make the user experience smooth: Cost might not be the main factor. An easy and familiar way to pay is crucial to get more sales. Keep in mind, a global outlook means changing how you do things in each market. When you cater to local payment likes, you'll open up a whole new world of chances. Source: Boku (Link in comments) #DigitalPayments #Fintech #Payments #Ecommerce #Cards

Subscription fraud is often invisible - but its impact is significant. Fake free trials and recurring payment abuse rarely appear fraudulent at the start. They typically mimic legitimate user behavior, making detection challenging. Common fraud patterns in subscription businesses • Multiple accounts created by the same user • Use of temporary emails and shared or stolen cards • Abnormal usage during trial periods • Intentional chargebacks after extensive consumption Business impact • Revenue leakage • Increased chargeback ratios • Payment gateway penalties • Distorted growth and retention metrics • Higher customer acquisition costs How fraud is detected effectively • Device and IP intelligence • Behavioral signal analysis • Payment reuse and failure patterns • Usage anomalies during trials and renewals Prevention strategies that scale • Limit free trials per device and payment method • Apply step-up verification for high-risk users • Monitor usage prior to renewals • Block bots and high-risk IP ranges • Leverage AI models to identify evolving fraud patterns Outcomes of a strong fraud strategy • Reduced fake users • Lower chargebacks • Accurate business metrics • Protected recurring revenue • Improved trust with genuine customers Fraud prevention is not friction. It is a safeguard for legitimate users and sustainable growth.

Founders focus on sales & marketing but often overlook what’s more important—actually closing the sale. The checkout experience. As e-commerce evolves, Indian customers have become pickier. They don't just want to buy your product—they want security, seamlessness, and an experience while doing it. Customers at the checkout page are the highest-intent customers, and losing them is unforgivable. Here’s what I learned while scaling Dr. Vaidya's by RPSG Group from 50 to 5,000 orders/day: 1.Fewer clicks and steps in checkout mean higher conversion. 2.When customers see a trusted payment interface, cart abandonment drops. 3.Single-click checkout leads to higher conversions, and multiple payment options help (card, UPI, BNPL, etc.). It doesn’t seem so important from the outset, but choosing a payment partner is crucial. When we switched to Razorpay, they checked the boxes: 1.Faster checkout, clean experience 2.Fewer abandoned carts 3.Multiple payment options Customers experienced a sense of familiarity, trust, and ease when interacting with it. And this led to real business outcomes. The e-commerce space is maturing in India, and operating a brand is now more science than art. Customers care, and the small things matter. Agree? PS: This shirt is my latest e-commerce purchase. It was a super clean checkout experience. Guess the brand and gateway :) #ecommerce #conversion #d2c #india #startup

Africa quietly processed 64 billion instant payment transactions worth nearly 2 trillion dollars in 2024. That is not a fintech headline. That is economic infrastructure hiding in plain sight. This week on the Unlocking Africa Podcast, I sat down with Sabine F. Mensah, Deputy CEO of AfricaNenda Foundation and co-author of the State of Inclusive Instant Payment Systems in Africa 2025 report, one of the most comprehensive studies ever produced on Africa’s real time payments ecosystem. What stood out most in this conversation was how clearly it reframed payments, not as a niche fintech topic, but as core economic infrastructure driving trade, productivity, and inclusion. As Sabine explained… “Digital payments mean more people are accessing and using digital payments and leveraging them to contribute to productive activities that can drive the economy.” Drawing on insights from 31 countries, we explored why Nigeria has emerged as Africa’s first fully mature instant payment system, and why this success was not accidental. In Sabine’s words… “It is not just about speed. It is about who is included and how systems are designed from day one.” We discussed: • Why scale alone does not guarantee inclusion • How interoperability transforms SME cash flow and liquidity • Why instant payments are foundational to AfCFTA success • How real time settlement changes growth outcomes for African businesses • Why trust, consumer protection, and recourse mechanisms matter as much as infrastructure One line that stayed with me throughout the episode… “There is no trade without payment. Digital payments are as important as ports and customs.” And a reminder that inclusion is deeply human... “It is not just one consumer with a bad experience. It is my family, my village, my community.” This episode is essential listening for policymakers, investors, founders, and anyone serious about doing business in Africa. Payment systems are no longer background infrastructure. They are central to growth. ⬇️ Listen now, link in the comments below ⬇️ #AfCFTA #DigitalPublicInfrastructure #PaymentsInfrastructure #AfricaTrade #InclusiveGrowth #Podcast