Finance systems need a new foundation as data models become outdated

As finance teams demand faster decisions and richer insights, experts argue that current accounting systems are too limited, pushing for a shift towards event-driven data models that enhance traceability and adaptability.

Finance systems were built to record what happened, not to explain what it meant. In a new essay for CFO Dive, Numeric argues that this is no longer good enough in an era when finance teams are expected to support faster decisions, richer reporting and AI tools that depend on context. The company says the general ledger remains useful for compliance, but too stripped back to serve as the main source of truth for analysis or automation. As a result, finance teams often rebuild business activity in spreadsheets, data warehouses and planning tools just to answer basic questions about performance. According to the essay, that duplication is not a minor inconvenience; it is a sign that the underlying data model is out of date.

The central complaint is that journal entries preserve only a narrow slice of the business story. They capture amounts and accounts, but not enough of the surrounding detail to show how a contract changed, which customer it involved, or how a transaction should be understood over time. The essay says that this creates three persistent problems: weak traceability, limited dimensionality and poor handling of time. Finance teams can struggle to move from a reported number back to the originating contract or business event, and they often have to rely on manual review to prove that the books are right. That makes variance analysis slow and makes later changes difficult to interpret.

The article uses revenue accounting to show how quickly the current approach breaks down. A contract may begin with an implementation fee and a subscription, then shift as seats are added, service levels are missed, discounts are granted, renewals are pulled forward or pricing assumptions are corrected. Each change adds another layer of complexity, yet the ledger records only the accounting outcome, not the evolving object behind it. The result, the essay argues, is that finance teams end up reversing entries, reposting values and maintaining side records of what actually happened. The accounting system, in this view, becomes a compressed archive rather than a living model of the business.

That is why the piece calls for a different foundation: one built on raw business events and the relationships between them. Numeric describes its Financial Data Platform as a system that stores source records first, applies accounting rules on top of that information and then generates outputs such as the trial balance downstream. The company says this approach allows the underlying data to remain auditable while also supporting more flexible reporting, such as revenue by geography and sales representative, or allocated cloud costs by customer. The claim is that accounting should become a derived layer, not the place where all useful information is lost.

The broader pitch reflects a wider shift in finance technology. Several newer platforms, including Numos, Pluvo, Numerus, Financiario and LedgerLM, are also positioning themselves around connected data, semantic layers, automating recurring analysis and making outputs more explainable. Their common argument is that AI in finance works best when systems can understand entities, timing and relationships rather than isolated entries. Numeric’s essay goes a step further by suggesting that accountants should increasingly act like finance engineers: defining policies, encoding exceptions and letting systems process the data automatically. In that model, the ledger still matters, but it is no longer the centre of gravity.