Amid the hype brought about by the advent of artificial intelligence in the Finance function, challenges concerning the technological advancement in sales and receivables forecasting remain.
According to Faisal Masood, director corporate strategy at Treasury Cube, unlike outflows, which can often be reasonably estimated based on contract terms, past payment behaviour, and fixed schedules, cash inflows—specifically from sales and receivables—are much harder to predict.
In an article, Masood enumerates factors that influence cash inflows, as they are not just a function of past trends:
- Macroeconomic Conditions: Inflation, interest rates, economic downturns, and government regulations.
- Industry-Specific Micro Factors: Customer behaviour, seasonal demand fluctuations, regulatory approval cycles, and competitive landscape.
- Business-Specific Internal Factors: Sales pipeline strength, marketing efforts, operational bottlenecks, and product pricing.
He explains that for any AI system to be credible in forecasting future cash flows, it must: backtest its models using real historical data; analyse correlation and causation of macro & microeconomic factors; continuously adjust factor weightages based on real-world results; and incorporate human feedback via Reinforcement Learning.
To be truly AI-driven, according to Masood, a cash forecasting solution must:
- Use macroeconomic indicators (interest rates, inflation, regulatory shifts) to dynamically adjust forecasts.
- Customise industry-specific models
- Backtest against historical external events (e.g., COVID-19 impact on sales).
- Refine predictions with user feedback (Reinforcement Learning).
- Be explainable—finance teams should understand how AI arrived at a given forecast.
