Optasia

The removal test leaves an unservable market, since the company's entire premise is that conventional lenders cannot profitably assess a five dollar advance to someone with no credit file. More than 200 machine learning models draw on thousands of alternative signals per user, producing roughly 300 credit decisions per second and around 1.5 billion a month, on a platform the company describes as designed from the ground up for markets. Its chief executive attributes the gap directly to modelling capability, noting that traditional banks do not come from the digital world and do not build algorithms of this kind.

No oversight mechanism is described, and at this scale none could exist in the conventional sense: roughly 300 credit decisions per second and 1.5 billion a month means every decision is made without a person, by necessity rather than by choice.

That is a defensible design for five dollar advances where manual review would cost more than the loan, and it makes the absence of any published description of thresholds, escalation, exception handling or post-decision review more consequential rather than less, because the controls that substitute for human judgement are entirely undisclosed.

The strongest evidence is structural rather than published: the company underwrites the default risk itself and provides guarantees to partner banks, so a model failure is its own loss rather than someone else's, and it reported a 36.2 million dollar profit with 76 percent revenue growth while carrying that exposure, which is a market test of model performance. The platform is patented and its architecture is described as purpose built for conditions.

Held at B because no accuracy, default rate, loss rate or validation result is published for a system making 1.5 billion decisions monthly, and a listed company carrying the credit risk has the figures.

This is the largest operational footprint in the index and it is corroborated across ratings agency, exchange and press sources rather than resting on company material. Distribution runs through 49 partners, principally mobile network and wallet operators including several of the largest African and Asian carriers, alongside 13 financial institutions across 38 countries.

Volume stands at roughly 1.5 billion credit decisions monthly, over 34 million transactions daily, more than 430 million annual active users and around six billion dollars of credit facilitated in a year, with over 20 billion extended since 2016. It listed on the Johannesburg exchange in November 2025 in that market's largest listing of the year, and a major South African banking group holds 26.1 percent.

No boundary statement was located. The platform sits between 49 distribution partners and 13 banks, several of which compete in the same markets, and its models improve by observing repayment across the entire network, which is the source of its advantage. Nothing states whether behaviour observed through one operator informs scoring on another's subscribers, what a partner contributes about its own customers, or what happens to a user's signal history when a partner relationship ends.

No data protection agreement, retention schedule, subprocessor list or consent framework was located, and the scale makes this the largest unaddressed privacy footprint in the index. Thousands of alternative signals per user are processed for over 430 million annual active users, sourced through mobile operators whose subscriber data covers communication, top-up and usage behaviour, across 38 jurisdictions with widely varying and in many cases recently introduced data protection law. Nothing describes lawful basis, what subscribers are told, or whether they can decline.

No attestation, certification, trust centre or enumerated framework was located. Thirteen banks and 49 telecommunications partners have completed supplier assessment before connecting the platform to funding and subscriber systems, and a public listing brings its own control expectations, so the underlying assurance is substantial and none of it is published.

Listing on a major exchange brings continuous disclosure obligations and external scrutiny that most vendors here do not face, a ratings agency has publicly assessed a transaction in its shares, and it operates through licensed banks and mobile operators in every market rather than alongside them.

One point deserves noting plainly rather than glossing: independent analysis observes that the platform model lets it scale across 38 countries without the capital burden and regulatory load of a balance sheet lender, so avoiding lending regulation is part of the structure's rationale. No financial regulator is named for its own activity in any market.

The access evidence is the strongest in this index and it comes with a mechanism and a comparison rather than an assertion. The chief executive states that the company serves customers with no formal credit history at risk levels lower than many retail banks, which is the two sided claim that distinguishes genuine improvement from loosened standards, and explains why banks cannot compete at a twenty or fifty dollar advance. About a third of volume reaches informal enterprises.

Against that, nothing is published on the consumer protection side of lending five dollar advances to very poor people at a rate of 1.5 billion decisions a month: no pricing, effective interest rate, rollover frequency, over-indebtedness monitoring or default consequence appears, and recovering loans through mobile data sales means the collection mechanism is tied to a service the borrower depends on.

No guarantee, indemnity or correction process was located for the borrower. The partner institution is unusually well protected, since the platform underwrites default risk and provides guarantees, meaning the vendor absorbs the loss when its models are wrong, which is rare in this index.

The consumer receives nothing equivalent: someone declined by a model reading thousands of signals from their phone use has no stated route to know why, to correct an input, or to contest a decision, and at 1.5 billion decisions a month no realistic appeal channel is described.

The data dependency is visible because it is the distribution channel: signals come through named mobile network and wallet operators whose subscriber behaviour feeds the models, so a buyer can see exactly what the scoring rests on and how concentrated it is. That transparency is a by-product of the business model rather than a disclosure decision, and it is real.

Held at B because the specific signal categories are described only as thousands of alternative elements, no data agreement terms with operators appear, and no model provider or infrastructure arrangement is identified.

The integration achievement is the business. Credit is embedded inside the mobile services customers already use, with several of the largest carriers and mobile wallet platforms in Africa and Asia named as live ecosystems, alongside 13 banks providing liquidity, across 38 countries. That means the platform is connected to telecommunications billing, wallet infrastructure and bank funding simultaneously, and repayment runs through mobile data sales as a collection channel. Reaching a subscriber inside their existing service, rather than asking them to visit anything, is what makes a five dollar loan viable at all.

No hosting provider, region selection or residency commitment was located. Operating across 38 countries while processing subscriber data from national mobile operators makes residency a live regulatory question in many of them, since telecommunications data localisation rules frequently apply independently of financial regulation, and none of it is addressed.

As a listed company it publishes audited financial disclosure well beyond anything else in this index, reporting a 36.2 million dollar full year profit for 2024, 76 percent revenue growth for 2025 that exceeded its own listing guidance, and a revenue mix showing microfinancing at 72 percent having overtaken the original airtime business. A ratings agency has assessed the shareholding transaction publicly.

Held at B rather than A because none of that discloses what a partner actually pays: no pricing, revenue share or commercial structure with distributors or banks is published, and for a platform earning from credit facilitated, the economics of the arrangement matter to any institution evaluating it.

Coverage spans 38 countries across three regions, with revenue distributed across Africa at roughly 65 percent, South Asia at 15 and the Middle East at 12, reaching banks, mobile network operators, mobile wallet providers and digital platforms simultaneously. End segments include consumers with no formal credit history and micro-enterprises, with about a third of credit volume drawn by informal businesses such as street vendors and shopkeepers. Products run from airtime advances through micro loans to working capital, and expansion into two further large African markets is under way.