Building a Forecast Treasury Can Trust

The first question treasury teams usually ask is about visibility. Where is cash today? What are the balances? What’s arriving by the end of week? Those are the right questions to start with, and Kyriba is built to answer them well. We process more than 3.5 billion transactions. Bank connectivity across hundreds of institutions. Visibility at the short-term horizon is something we’ve invested in deeply, and it shows.

What we have also been building over the past several years is the next layer: the indirect method, and the modeling logic needed to construct a forecast instead of simply reporting one. FP&A is a vast world, and I will not pretend treasury has a simple, one-size-fits-all path into it. But we are clearly on that journey. More and more, clients are asking how to move from FP&A assumptions to cash impact, and that is exactly where liquidity planning needs to become stronger.

But here’s something I’ve observed over six years working specifically on liquidity: the teams with the strongest short-term cash processes eventually hit the same friction point. It can happen after a few weeks. It becomes very clear once you get into the 13-week horizon. The data changes. The sources change. The level of certainty changes. And often, the process for handling all of that has not kept pace.

Why long-term cash forecasting is different from short-term cash management

The further out you go, the more the forecast changes. You are no longer relying only on bank balances, transactions, and confirmed cash flows. You are working with ERP data, open invoices, accounting entries, open orders, FP&A budgets, payment terms, planning assumptions, and business rules.

Same company, very different forecast.

That is why I describe long-term liquidity forecasting as a construction exercise, not a reporting exercise. Treasury is not just aggregating what already exists. The team is building a model of what is expected to happen, based on logic that has to be applied consistently across thousands of inputs.

Capability	Cash Management	Advanced Liquidity Planning
Primary focus	Today's cash position and near-term visibility	Long-term liquidity forecasting and planning
Time horizon	Days to weeks	Months to 24 months
Main inputs	Bank balances, transactions, short-term cash flows	ERP data, invoices, open orders, budgets, assumptions
Core value	Visibility and control	Forecast construction, scenarios, versioning, decision support
AI readiness	Helps establish connected data	Provides governed logic and planning structure for AI-guided forecasting

Why Excel-based liquidity forecasting breaks down at scale

I want to be concrete about this, because “Excel doesn’t scale” is easy to say and rarely useful on its own.

Take the direct forecasting method. You have invoices in your ERP. From those invoices you can derive expected inflows and outflows. That seems manageable. Then you get to tax. In Europe, that means VAT: extracted separately from the invoice value, calculated, and positioned at the correct moment in time. In the U.S., you’re handling sales tax, with its own timing logic. The tax cash flow doesn’t land on the same date as the invoice. It’s a different number, positioned at a different point in time.

Then add the AP side. You may know your invoice due dates. But you also control when you actually pay. If your company runs vendor payment campaigns on the 15th of each month, the actual cash outflow is the result of aggregating all the invoices due in that window and shifting that total to the payment run date. It’s not a formula. It’s a sequenced calculation.

You can work around this in Excel with enough tabs and enough expertise. I’ve heard treasury teams describe exactly how they do it, and I believe them. But one treasurer put it plainly: “The next three to five years, we have a huge pipeline of work, and it’s not sustainable in Excel.” Even someone very crafty can’t model all of that consistently, especially at volume, especially when the model lives in one person’s head.

That’s where you need a calculation engine, not a spreadsheet.

What Advanced Liquidity Planning does

When we designed Advanced Liquidity Planning, the goal was not to build a better reporting view. Treasury already has plenty of places to look at numbers.

The word I use internally is workspace. A workspace is where you make things. Liquidity planning becomes a workspace where treasury has a growing set of capabilities to construct forecasts: define the rules, apply the business logic, pull in invoices, open orders, budgets, business calendars. Then bring that constructed plan back into the cash management view teams already trust and use every day.

That framing matters because it changes what the tool is for. It’s not for seeing more data. It’s for building something useful from the data you already have.

How the Unified Worksheet connects short-term cash and long-term liquidity planning

The handoff between real-time cash management and the constructed long-term plan came up constantly in customer conversations, especially during implementations. Teams would ask: how do I make the junction? How do I connect what I’m doing in cash management with what I’m building in liquidity planning?

The Unified Worksheet answers that. We draw a line. On one side is the rolling forecast world: updated balances, confirmed transactions, and short-term cash flows. On the other side is the version-based planning world: the liquidity plan constructed from budgets, assumptions, rules, and models.

The Unified Worksheet brings both together at a cut-off date, creating a single view from today through the next 12 months or more. One workspace, one story to bring to the CFO, instead of stitching together a daily positioning report, a separate forecast file, and a budget extract and calling the result a liquidity view.

How modeling feeds the long-term liquidity forecast

Once the Unified Worksheet connects short-term cash management with long-term liquidity planning, modeling is what feeds the longer-term view. For a 13-week forecast, that may mean turning AP and invoice data into expected cash flows, applying payment terms, tax timing, payment campaigns, and business-day rules.

For a 12-month forecast, it may mean translating a P&L budget into liquidity flows by applying timing assumptions, seasonality, and business logic. And when planning starts in a dedicated FP&A platform, Kyriba can connect that plan to liquidity planning rather than asking treasury to rebuild it somewhere else. In every case, the goal is the same: turn business data and assumptions into a forecast treasury can govern, explain, and trust.

Why AI-powered forecasting requires a strong data foundation first

We announced our Agentic AI solution at KyribaLive, and I’m genuinely excited about where it’s going. In liquidity planning, AI already points to a practical shift: using historical cash flow and liquidity data to help generate forecasted flows, while keeping the forecast anchored in configured data sources and planning structure.

But I’ll say the quiet part out loud: AI is not going to rescue a poorly structured forecasting process. If assumptions are unclear, if the rules are buried in spreadsheets, if there’s no version history and no governance, AI will accelerate the confusion rather than fix it.

One treasurer put the aspiration exactly right: “What I’d like is for the person who’s forecasting to be the reviewer, with AI as the engine that actually does the forecast.” That’s the direction this is heading. But it requires the foundation to be solid first. Advanced Liquidity Planning builds that foundation. AI-powered forecasting builds on top of it. That’s the right order.

A stronger foundation for better liquidity decisions

For me, Advanced Liquidity Planning comes back to one simple idea: treasury should spend less time assembling the forecast and more time using it. When the data, rules, assumptions, and versions are connected, the forecast becomes easier to explain and easier to trust. That is what gives treasury more confidence when the business asks the next question.

See the demo