By Simone Vaccari, Chief Credit Officer, Bboxx
A common language for capital
Standardized metrics are no longer just a reporting exercise in this sector. They have become core infrastructure, and that shift is both necessary and overdue. Capital moves faster when everyone is speaking the same language, and PAYGo has needed that language for years. That was the point I made as a panelist on GOGLA’s recent webinar on the revamped PERFORM KPI framework.
That point carries a second half, though. Infrastructure is not the whole building, because PERFORM was built for covenant reporting to lenders, deliberately narrow rather than an attempt to standardize every credit metric, and that scope was the right call. My point is simply that the real risk in this business still lives outside what that scope was ever meant to cover.
Investors want comparability, and they are right to want it. Diligence across a dozen portfolios in a dozen markets is slow and expensive when every operator defines repayment a little differently. What I want to add to that conversation is a caution that comes from actually running a credit book across several African markets, where the texture of risk shifts more than any standardized indicator can capture on its own.
What the numbers actually show
The revamped PERFORM framework centers on two anchor metrics, repayment rates and customer ownership rates. Both are sensible choices, and both are metrics we already track closely at Bboxx. What I wanted the audience to take from that session was less about the metrics themselves and more about the level at which you choose to look at them.
We made the shift to cohort-level analysis internally back in 2024, and it remains one of the more useful changes we have made to how we read our own book. A portfolio-level view can look healthy simply because a business is growing quickly, since new loans dilute the visible effect of older loans going bad. Look at the same book cohort by cohort, tracking each vintage of customers on its own curve, and problems that were hidden in the aggregate tend to surface months earlier.
At Bboxx, since we made that change in 2024, the conversation with investors has shifted along with it. Where most of our calls used to start with a question about how we calculate a given number, those same calls now tend to start with a question about what the number actually means for performance and risk, which is a far more useful place to spend an hour.
There is a second layer worth holding alongside the cohort view, and that is timing. PAYGo repayment behavior moves with the rhythm of household income, which in many of our markets means it moves with harvest cycles, with school fee seasons, and with other periodic swings in what a family has available to spend in a given month. A standardized number taken on its own, without that context, can look like a warning sign when it is really just the calendar doing what the calendar always does. We have learned to read our PERFORM numbers alongside a short seasonal note, so a dip in a harvest off-month reads as exactly that, rather than as a portfolio quietly going wrong.
What does not show up cleanly in any standardized framework is the risk that is specific to a market rather than to a business model. Fraud, early termination, and device tampering behave differently from country to country, shaped by local incentives, local enforcement, and even the secondhand market for the hardware itself. We see fraud patterns that are simply more pronounced in some markets than in others, for reasons that have more to do with local conditions than with anything a repayment ratio on its own will ever tell you.
That is not a criticism of the framework. It was built for covenant reporting and cross-market comparability, and it does that job well. It was never meant to be a diagnostic tool for the kind of localized erosion that shows up in a specific city or a specific agent network. That is why we run our own layer underneath it, including operational measures as simple as the share of customers whose collection rate falls below 30 percent by a defined point in their loan term. That number will not appear in any investor deck built around PERFORM, but it is often the first place we see trouble start.
What this means in practice
Standardized KPIs will cut diligence time for investors, and they should give real confidence when comparing across portfolios. But the more important message here is for operators. Adopting these metrics is necessary, not sufficient, and treating them as enough to run the business well is exactly where the risk creeps back in.
Internally, our approach has been to keep the metrics we report under PERFORM stable, treating them as the shared language for our board and for investors, while keeping a standing local risk lens in every management pack. That lens covers fraud, early termination, device patterns, and whatever else does not travel well across borders.
The next advantage
Standardization is quickly becoming table stakes rather than a differentiator, and that is exactly as it should be for an industry trying to grow up. The operators with an edge from here will not simply be the ones with the cleanest PERFORM numbers, since most credible operators will get there eventually. The edge will belong to whoever pairs that clean, comparable core with sharp, local diagnostics built to catch what the standardized view was never designed to see.
PERFORM gives this sector a common language, and that matters more than people arguing over KPI definitions might admit. But the real work of managing credit risk in PAYGo still happens in the margins that refuse to standardize, market by market, cohort by cohort. That is where I intend to keep looking, and it is where I would encourage any operator working in this sector to keep looking too.
