What Five Days in the Philippines Taught Leaders About the Future of Digital Financial Services

A week inside Manila’s digital economy revealed why the most consequential platform decisions are shaped by what leaders see firsthand.

Whether you’re an operator sequencing a product launch, an investor sizing up a market entry, or a policymaker choosing what to regulate first and what to leave to competition, seeing how decisions play out in-market changes the quality of the decisions that follow. A report can show that a platform has been adopted, but it cannot show what happens to the money, the customer, or the algorithm once that platform goes live.

Caribou recently led a Live Learning trip through Manila, Philippines. The group came from across Africa with specific questions. How does a platform business move from payments penetration to genuine financial participation? What does the build-or-partner decision look like when operating at scale? 

Four conversations on the trip, on payments infrastructure, platform strategy, AI-enabled credit, and informal supply chain credit, arrived at the same lesson: what looks obvious from a boardroom looks different from inside the market. 

Digital retail transactions in the Philippines grew from 1% of all purchases in 2013 to 57.4% in 2024, a jump that is not surprising given Southeast Asia’s shift to digital over the same period. The easy explanation is COVID. The pandemic pushed people online, and lockdowns closed cash-only stores, but the groundwork that made the shift away from cash possible was laid a decade earlier, long before anyone had heard of coronavirus.

In 2015, in order to move the Philippines from an economy where 99% of transactions still ran on cash, and where no commercial incentive existed for competing banks and fintechs to connect their systems to each other, the Philippines central bank made an unusual choice. Rather than waiting for banks and fintechs to cooperate, it required them to. It built shared digital payment rails linking every bank account and e-wallet, and by 2023 mandated a single QR code standard so a shopper could scan the same code in any store, with any app, nationwide. The infrastructure was in place years before most people started using it. 

Two national programs closed the gap between infrastructure and use: the government’s 4Ps welfare payments, reaching 4.4 million households, and remittances from Overseas Filipino Workers (OFWs), both routed money directly into digital accounts, so millions opened wallets by default. 

Before this Live Learning trip, the traveling team’s working assumption was that, with the Philippines’ interoperable payment rails, a merchant with a working QR code could be counted as digitally included. On the trip, a covered market in Pasig City invalidated that assumption. We watched sari-sari store owners accept QR payments from GCash and Maya, then turn the phone screen toward the customer to confirm the sale, not for their own benefit but the customer’s. Their wholesale suppliers ran on cash, so any digital income was cashed out immediately. 

This pattern is not new: Ignacio Mas argued in 2013 that mobile money had made cash more efficient, not replaced it, and Caribou’s own 2017 research named this gap “effective use,” the distance between holding an account and putting digital value to work.

Account ownership and financial inclusion are distinct outcomes, and the Philippines has achieved the first without finishing the second. While 85% of households have access to at least one account, individual adult account ownership has stalled near 50% since 2021, and only around 30% of adults can meaningfully absorb a financial shock. Standing in Pasig City showed what the data did not: adoption and participation can diverge. Adoption figures will not show this gap. A market visit will. 

Over the trip, we split time across three of Manila’s biggest platforms, each of which made a different bet on where platform value lives.

By the time we sat down with GCash near the end of the Live Learning trip, every operator, investor, and regulator we had met that week had described GCash’s journey to scale in near-identical terms: a convergence no single report could show, visible only across the three visits in sequence.

The assumption most leadership teams bring into a market like the Philippines is that one model eventually wins. Three days in the Philippines showed that three structurally different bets can all be credible simultaneously. 

A third thread of the trip convened five practitioners from different layers of the Philippine credit ecosystem. The leadership team assumed that AI had moved the frontier. The panel produced was a more foundational picture: a detailed inventory of what has to exist before AI in credit is significantly useful at all. The infrastructure challenge must be solved first. 

Kasper Svendsen of Datung.io extends credit to tricycle drivers and sari-sari store owners whose transaction data tells only part of the story. The model is AI-driven, but the credit enhancement that makes it work is not. Datung.io lends through community groups, where the social cost of being seen to default by people you live among functions as collateral. A discussion originally intended to explore what AI had made possible in Philippine credit uncovered that the most effective risk-mitigation mechanism turned out to be peer accountability.

Every other practitioner arrived at the same structure from a different angle:

The panel opened as a conversation about what AI had made possible in Philippine credit. By mid-morning, the unexpected insight was more about what had to exist before any of it could work. None of the AI on display in Manila preceded its foundation: an identity layer, a data network, a cost structure, or a social network built over years, and that principle travels far beyond the Philippines.

In the Pasig market we visited, the most effective B2B lender was neither GCash nor Maya, but the community leader who keeps a paper ledger and has extended credit to the same stall owners since before most of them had a QR code. She knows who pays on time and whose stall has run in the family for a generation, and prices credit accordingly. Her rate is built into the markup, and her security is social obligation and twenty years of local knowledge. Two decades of fintech investment have not displaced her.

The technology reached her stalls: the QR codes are real, the GCash wallets are active, but the legal infrastructure that would let formal credit compete on her terms has not. Maya extends merchant credit at rates that represent a real improvement on the informal lending that preceded it. One merchant we met carried a US$900 credit limit through GCash, a true improvement on the informal rate. But both lines are underwritten to the individual running the stall, not the business, so a lender sees a person’s transaction history, not the cash flows, inventory cycles, and supplier relationships the wholesaler already tracks from twenty years of watching. Until a business is formally separable from the person who started it, formal credit cannot underwrite it, or price low enough to compete with her.

The AI credit panel earlier in the week focused on the same underlying problem, but one level down. Boost Capital, LenderLink, and BillEase have each spent years building the infrastructure that lets a digital lender know, with confidence, who an individual borrower actually is. Building the equivalent for businesses is the next round of the same work. Three things need to be in place before formal credit can displace the informal wholesaler: suppliers willing to accept digital settlement, business lending priced low enough to undercut her markup, and platforms that can clear fast enough to match her flexible terms. The sequence that worked for consumers has to be built again, one level up.

The Philippines built payment rails before it required anyone to use them. Three of its platforms are proving that more than one growth model can win in the same market. Its credit practitioners are proving that AI underwriting is only as good as the identity, data, cost, and trust infrastructure underneath it. Its most effective merchant credit still runs on a paper ledger because no lender can yet separate a business from the person who runs it. 

The product stack, the scale of OFW remittances, and the shape of Manila’s credit market are particular to the Philippines and should not be mistaken for a template. What is transferable is the sequencing logic itself, visible only at street level and local convenings that leave plenty of room for live discussion. Adoption dashboards show part of the picture. A market visit shows it in full.

The same principle applies elsewhere, but the questions each market answers are entirely different. In China, the questions are different. They are about AI, super-apps, manufacturing at scale, and what it actually looks like when an entire ecosystem, not just a company, is operating at full speed. In Dubai, the lessons are different again: the market surfaces insights on regulation, innovation, public-private collaboration, and how policy and market design can accelerate a market on purpose, rather than waiting for one to emerge.

The role of Live Learning is matching the right market to the questions your leadership team needs to answer.

—
Want to see and understand markets like this for yourself? Caribou’s Live Learning practice designs immersive market experiences for operators, investors, and policymakers navigating exactly these sequencing decisions. Explore Live Learning, or reach out to Shirley Gilbey to start a conversation.