06-reference

mostly metrics rolling 13 week cash flow claude excel

2026-07-02·reference·source: Mostly Metrics·by CJ Gustafson
cash-flowfinancial-modelingclaude-excelworking-capitalcfo-tools

"How to Build a Rolling 13 Week Cash Flow Forecast Using Claude + Excel" — Mostly Metrics

Why this is in the vault

Tutorial #4 in CJ's Claude + Excel series. Directly relevant to RDCO's AI COO stack: demonstrates a practitioner workflow for synthesizing messy multi-source data (8 tabs) into a live cash model using Claude as the intelligence layer. The core tension — P&L profitability vs. cash-poor working capital — is a recurring CFO failure mode that shows up in every growth-stage B2B company. This is exactly the kind of financial operations content the AI COO layer should know cold.

The core argument

Net 30 payment terms are contractual fiction. The real pattern at most B2B companies — especially those with Fortune 500 or enterprise customers — is Net 55-70 in practice. This gap between stated terms and actual collections behavior is where cash forecasts fail and where covenant-breach risk hides.

CJ's workflow for Meridian Software (a mock company he built for the exercise):

Eight data sources synthesized:

  1. Bank account statements
  2. Payroll files
  3. AR open invoices
  4. AR collections history (how does this customer actually behave?)
  5. AP vendor schedules
  6. Misc. outflows (D&O insurance, team offsites, one-off commitments)
  7. Venture debt covenant thresholds (minimum cash balance triggers)
  8. Deal pipeline / expected close timing

What AI adds here: Claude can synthesize data points that contradict each other across sources — e.g., an AR invoice says Net 30 but the collections history shows the customer has paid in 55-70 days for the last 8 quarters — and build a realistic timing model rather than trusting the invoice terms.

Paywall note: The free portion is an intro/teaser. Paid subscribers get the full build video, exact Claude prompts, the raw 8-tab data file, and the finished model. The framework and mental model are fully accessible in the free portion.

Quote that anchors the thesis:

"Net 30 is a lie, and yet we keep telling it to our forecast."

This is the CFO version of "garbage in, garbage out" — except the garbage is optimistic contract language that no enterprise customer has honored in years.

Mapping against Ray Data Co

High relevance. Three direct connections:

  1. AI COO financial ops: The multi-source synthesis pattern (bank + AR + AP + payroll + misc) is precisely the kind of structured data-wrangling that an AI COO should automate. If RDCO ever builds a cash dashboard or financial ops layer, this is the reference workflow.

  2. Claude + Excel as a workflow pattern: This is tutorial #4 in a series CJ is building. RDCO should track this series — each installment is an incremental building block for finance-workflow automation using Claude. The prior tutorial (sales capacity model) + this one (13-week cash) are the two highest-value installments so far for a DSA/CFO audience.

  3. phData context: Deal Solutions Architects often help clients instrument their financial data stack. The collections-timing DSO analysis (actual vs. contractual pay behavior) is a pattern that maps onto Snowflake data pipelines for AR analytics — a legitimate discovery/scoping angle.

Tactical note: The 13-week rolling horizon is the canonical CFO liquidity window. Any AI-assisted FP&A product targeting the CFO buyer needs this as a core use case, not a stretch.

Related