The Hidden Cost of Legacy SaaS in African Insurance

The administration platforms that dominate South African insurance — the high-volume retail and funeral cover incumbents whose names every broker recognises — market themselves heavily on the language of automation. The marketing is not, strictly speaking, false. The platforms do automate. But they automate within a window so narrow that the automation collapses on contact with anything resembling complexity.

These systems operate on manually configurable product rules. Waiting periods, grace periods, premium escalations, age-banding — all of it is deterministic, all of it expressed as if-then logic, all of it dependent on the input being clean and structured. When a claim arrives that contains unstructured data — a handwritten medical note, a witness statement that contradicts a police report, a photograph whose metadata has been stripped — the platform has no internal mechanism to resolve the complexity. It flags the claim. It routes the file to a human queue. The automation, at the moment automation would actually be valuable, ceases.

The generic payment aggregators are a different category of incumbent and a different category of failure. They process billions in monthly bill payments across vast retail networks, and on the strength of that financial plumbing they have begun offering simple insurance products. As pipes for moving capital, they are unparalleled. As insurance administrators, they are crippled by the rigidity that made them excellent payment switches in the first place.

The aggregators treat an insurance premium exactly as they treat a prepaid electricity top-up: as a binary financial transaction. They cannot natively ingest a complex claims dossier. They cannot evaluate property-damage nuances. They cannot dynamically negotiate a payout. The architecture has been optimised for the move, not for the decision. When a deviation arises, the aggregator does what the policy administration platform does — it escalates to a human.

The legacy platforms are not failing in the sense of crashing. They are failing in the more pernicious sense: they perform exactly the function they were designed to perform, but the function itself is no longer the function the industry needs. They are systems of record in an environment that demands systems of action. The next instalment examines what a system of action actually is, and why the difference is architecturally absolute rather than incremental.