Forecast renewals from the contract record. Not the spreadsheet.
Most SaaS finance teams rebuild the renewals model from scratch every quarter. They pull from Salesforce for the dates, NetSuite for the dollars, the CSM platform for the health score, and the head of the analyst who has been there longest for the rules. The number that lands in the board pack is the best anyone could do under deadline. It is also untraceable — when the actual variance arrives, nobody can recover what was believed about a given contract three months ago, because no one wrote it down.
HarborOS treats the renewal forecast as a derived artifact, not a manual rebuild. The forecast reads off the contract record in Contracts, applies the rules captured in Lighthouse, and produces a versioned snapshot every week. Every dollar of movement traces to a specific contract decision. Every assumption survives turnover.
It is built for the weekly operating rhythm of a PE-backed software CFO — not for the once-a-quarter scramble.
The renewals model is the most consequential spreadsheet in the company. It is also the least owned.
The contract lives in a PDF in a shared drive. The renewal date lives in Salesforce, entered by whoever closed the deal. The invoice amount lives in NetSuite, in a currency that may or may not match the contract. The customer health score lives in a CSM platform that nobody on the finance team logs into. None of these systems agree on what counts as a customer, what counts as ARR, or when a contract is considered renewed.
The VP of Finance carries the rules in their head. Auto-renewals roll forward unless we hear otherwise. Professional services are walk-in run rate — they don't count in the base. Multi-year contracts amortize at the booking value, not the ramp. The rules are sound. They are also undocumented. When the analyst who has been there longest leaves, the rules leave with them.
When the CFO asks for next quarter's forecast, someone opens a blank Excel file. Two days later, a number appears. It is plausible. It is the best anyone could do. It is also unauditable — there is no way to ask, six weeks later, what the model assumed about Acme on a Tuesday in February. The assumption was never written down. Only the number was.
The result is a forecast that drifts by ten to fifteen percent through the quarter, a board pack assembled the night before, and a CFO who cannot defend the underlying judgment without flipping to a sheet they have not seen since the file was last opened. The number is real. The defense is not.
The forecast is a query, not a rebuild.
HarborOS gives the renewal forecast a defined data source, a defined ruleset, and a defined output. The forecast is recomputed every time a contract changes. It is locked, dated, and comparable from week to week. The model and the assumptions live in the same place.
The contract is the source of truth. Every dollar in the forecast traces to a specific row in Contracts. Term dates, ARR value, billing entity, currency, renewal terms, and auto-renewal clauses are captured once on the contract record and read by every module downstream. There is no separate renewals tab. There is no second source. When a contract is updated in Contracts, the forecast reflects it within seconds.
Your rules are written down, not held in someone's head. The rules that used to live in the analyst's head — how auto-renewals are interpreted, what counts as walk-in run rate, when a contract is "at risk" — are captured in Lighthouse as plain-language house rules. The forecast applies them by construction. A new analyst can read the rules on a Monday and produce a defensible forecast by Friday.
The forecast updates on every contract change. When a contract is added, modified, or stamped in Docking, Compass recomputes. The model is not rebuilt — it is read again. The team works the call sheet in Portolan; the forecast reflects each move as it happens. The Monday-morning number is current as of Friday-evening lock, not as of three weeks ago.
Every snapshot is locked and comparable.When the CFO commits a forecast in Compass, it becomes a Snapshot — immutable, dated, signed. Next week's snapshot can be diffed against this one, contract by contract. Six months later, the February snapshot is still exactly what was reported in February, not a reconstruction. The audit trail PE firms have always wanted is the artifact, not a side document.
What changes when the model derives from the contract record.
| Before HarborOSSpreadsheet, rebuilt quarterly. | With HarborOSQuery, recomputed weekly. |
|---|---|
| Renewal forecast rebuilt quarterly in a 14-tab Excel file. The model is a snapshot of two days of work. | The forecast derives from the contract record, every week. The model is a read against a live dataset. |
| ARR definition lives in the VP of Finance's head. Auto-renewal logic lives in the analyst's notebook. | The rules live in Lighthouse. Written once, applied always, readable by anyone on the team. |
| Variance between forecast and actual takes a week to explain — and the explanation never reconciles to the dollar. | Every dollar of variance traces to a specific contract decision in the snapshot diff. The explanation is the data. |
| Forecast accuracy drifts to ±10–15% by the end of the quarter. Nobody can say which contracts caused the drift. | Accuracy holds inside ±2–3% by the second week of the quarter — because the contract movements are visible week-over-week. |
| The model and the assumptions live in different files. Neither is versioned. | The model is the assumptions. Both are locked together in each Snapshot. |
| Onboarding a new analyst takes a quarter. The rules have to be re-learned from oral tradition. | The rules are written. A new analyst is contributing to the forecast by Friday of week one. |
Built around the weekly finance rhythm.
The renewal forecast is not a quarterly artifact. In a PE-backed software company, it is a weekly artifact — reviewed Monday, worked Tuesday through Thursday, locked Friday. HarborOS is built around that cadence. Every surface in the product is designed to be touched on a specific day.
Friday's locked snapshot lands in the inbox. The CFO reads the auto-generated variance narrative in under five minutes — the diff against the prior week is already written.
The renewals team opens Portolan. Approaching renewals are sorted by date and confidence. Contracts move through the pipeline in real time.
Updates land back in Contracts. Stamped records appear in Compass instantly. The forecast number recomputes without a rebuild.
Variance review with the leadership team. What changed since Monday? The answer is a contract-level list, not a slide deck.
The CFO reviews, commits, locks. The Friday snapshot is the artifact — immutable, dated, signed. It becomes Monday's variance baseline.
A renewal forecast worth defending.
Tied to the contract record.
Updated every week.
Locked, dated, and comparable.
Built for the board pack — not for last Thursday.