Case Study: A 13-Week Cash-Flow Forecast, Built Live Against NetSuite

About This Case Study: Client name, people, and referral partner have been withheld. All claims trace to engagement records, and the numbers are real model parameters from the build unless marked illustrative. No invented ROI figures appear in this article.

The Client

Client: A mid-size North American manufacturer of specialty power-electronics components. Their customers are OEMs, utilities, data centers, and industrial operators — large orders, long lead times, and a cash cycle dominated by receivables. ERP: NetSuite.

Who used it: The company's fractional CFO. Not an analyst handed a tool by IT — the finance lead herself, working directly against live NetSuite data in weekly working sessions.

The Problem

Like most mid-size manufacturers, their cash position lived in NetSuite — spread across open sales orders, AR aging, and history — but their cash forecast lived in a spreadsheet someone had to rebuild. Two specifics made it hard:

• Order timing is unreliable at the source. Sales reps don't update ship dates on sales orders when they slip, so any forecast that trusts the typed date inherits the slippage.
• Aggregate aging buckets lie. A large OEM that pays Net-45 on the dot and a slow-paying industrial account shouldn't be averaged into one collection curve.

What Was Built

Over a series of weekly working sessions, the CFO and an NSGPT agent built a rolling cash-flow forecast directly on the company's NetSuite data. A 13-week rolling view is the primary horizon, with a monthly view extending six-plus months.

• 13 weeks — Primary Rolling Horizon
• 3 layers — Orders, AR, Seasonality
• 6+ months — Extended Monthly View

The model has three layers:

1. Open sales orders → cash. Each open SO is converted to expected cash using status-dependent logic — an order pending approval is treated differently from one pending fulfillment or pending billing — combined with ship lead time and collection timing.
2. Open AR → collection timing. Aging buckets carry collection-probability curves, with customer-level overrides driven by each customer's actual payment history rather than the bucket average.
3. Seasonality and pipeline. Beyond the order book, the longer horizon blends historical seasonality, known pipeline, and manual overrides.

The model parameters give a feel for the mechanics: status haircuts on the order of 70% for pending approval up to 98% for pending billing, and collection curves stepping down from ~98% on current AR to ~40% at 90+ days past due. (These are the engagement's working parameters — every deployment tunes its own.)

The build itself was iterative in a way a static report never is. Mid-engagement, the CFO changed the methodology: instead of haircutting revenue amounts by probability, the model shifted to extending lead times by order status — assume the revenue lands, model when. That's a structural change to the forecast, made in conversation, against live data.

The ship-date problem got an explicit treatment too: rather than trusting typed dates, the model can apply historical slip factors by product line or a rolling actual-lead-time average.

The Revenue Layer the CFO Built Herself

The part worth underlining: the sales-order inflow analysis underneath the forecast — the revenue layer — was built by the CFO herself between sessions, in plain English, against NetSuite. During the same period she ran a full month-end close through the platform, converting twelve of her standing close procedures into NSGPT workflows.

Her comparison point was general-purpose AI chat tools she'd used for building spreadsheet formulas. The difference she called out was collaborative dashboarding: output that lives as a shared, refreshable artifact on the company's real data, not a formula pasted back into Excel.

The Auditability Requirement: She also pushed on something we think every finance team should demand: auditability. A CFO won't sign off on a number she can't trace, so the methodology has to be readable in plain English alongside the code that computes it. That requirement is now part of how we build.

Where It Stands

A working v1 of the three-layer forecast, in active weekly use and refinement by the finance lead, ahead of an executive review for broader rollout. No invented ROI numbers here — the honest claim is narrower and, we think, more interesting: a fractional CFO built and now operates a structurally sophisticated, customer-level cash-flow forecast on live NetSuite data, without a BI team and without an implementation project.

Names withheld. All claims trace to engagement records; numbers are the engagement's working model parameters unless marked illustrative.