Traditional forecasting models and predictive analytics are breaking down under the weight of unprecedented variability, whether from tariffs, geopolitical shocks or macro dynamism. The ongoing strain on old data is forcing forward-thinking chief financial officers to turn to a new tool: data that doesn’t exist yet.
Synthetic data, or artificially generated data that mimics the statistical properties of real-world information, is gaining favor among predictive analytics teams in B2B organizations. Once a niche tool in academic artificial intelligence research and spear-tip areas like self-driving car simulations, synthetic data has crossed over into finance, where it is helping to reshape areas like treasury strategy, liquidity planning and supply chain resilience.
B2B companies, challenged by lumpy revenue patterns, seasonal working capital cycles, and limited historical visibility in new markets, are especially ripe for this shift. Unlike consumer-facing firms, which can draw on vast lakes of behavioral data, B2B finance teams often lack the volume and granularity needed to build resilient predictive models.
For example, finance teams can turn to synthetic data pipelines to simulate market swings, anticipate supplier default scenarios, and visualize cash flow risks before they materialize. These scenario planning capabilities are commonly repeatable and scalable.
Read also: Payment Data Is Back-Office Automation’s Hidden Gold Mine
Advertisement: Scroll to Continue
Synthetic Data Moves From Silicon Valley to the CFO Suite
Synthetic data was initially popularized by AI labs and tech firms using it to train computer vision systems and natural language models. But as the uncertain operating landscape puts new pressures on traditional forecasting methods, demand for alternative data sources is spilling over into corporate finance.
By programmatically generating thousands of plausible but non-identical datasets — each aligned with real-world statistical characteristics — companies can simulate how their businesses might perform under high-stress conditions. These include foreign exchange shocks, supplier shutdowns, port congestions and cyberattacks.
“[Accounts receivable (AR)] is no longer about settling the past. It’s about predicting the future of cash,” Pamela Novoa Ralli, head of product management at FIS, told PYMNTS in an interview published Tuesday (Aug. 5). “It’s moving from a responsive to a proactive view.”
“AI allows predictability features to be at the core of the AR solution, instead of what we have seen in the last 10 years, which is about looking at the past and focusing on efficiency and effectiveness of the current state,” she added.
For treasury teams managing billions of dollars in cash, even small blind spots in data visibility can have large consequences. This is where synthetic data may end up proving transformative.
“There’s nothing more important to the treasurer than preserving and understanding where their cash is,” Tom Durkin, global product head of CashPro in Global Payments Solutions at Bank of America, told PYMNTS in an interview published Monday (Aug. 4), adding that people traditionally have used a spreadsheet to forecast, and that could take up to a week to complete.
By the time it’s done, the data is out of date.
The PYMNTS Intelligence report “Why Treasurers’ Influence Matters” found that treasurers with high levels of influence are more likely to report that their companies have predictable cash flows, expect revenue to increase and are agile in responding to shifting market conditions.
See also: With Trade War Back on, CFOs and Product Chiefs Reset and Reboot
Accelerating the Shift From Reactive to Predictive Finance
The ripple effects of synthetic data are also poised to reshape how CFOs collaborate with operations and procurement leaders. Amid rising supply chain fragility, finance teams are tasked with not just tracking costs — but actively shaping the network architecture.
Synthetic data, for example, also allows firms to test the resilience of just-in-time and just-in-case inventory models across divergent futures — something historical data, by definition, cannot do.
In regulated sectors, synthetic data also helps avoid data privacy risks. Because the data doesn’t map to real individuals or firms, it can be used to train AI models without violating GDPR, HIPAA or other data protection laws — making it especially attractive for financial services and healthcare firms.
But while synthetic data promises great upside, it also has challenges. The primary concern is fidelity. If the generative models are poorly trained, they can produce unrealistic or misleading outputs.
For all PYMNTS B2B coverage, subscribe to the daily B2B Newsletter.
