Planning that reconciles itself to reality.
Budgets, forecasts, scenarios and cash sit on the same ledger as your actuals. Variance is computed rather than compiled, the forecast re-fits to actuals in one run when a period closes, and there is no version-17-final spreadsheet to rebuild.
One engine. Every plan you’ll build on it.
Budget, forecast, scenario, cash, labor and department spend are models on one dimensional, time-series, driver engine. Not five tools that then have to agree with each other.
One coordinate system
A plan line sits at account × entity or dimension × period, the same coordinates the ledger posts to. So a budget slices exactly the way your books slice, by entity, department, project, location or fund, and a plan for one entity is that entity’s plan rather than a share of a company total.
- Budget, forecast and scenario versions on one engine
- Plan lines at account × dimension × period
- Duplicate any version to fork a scenario, leaving the original untouched
Budgets you seed, then argue about
Seeding takes the same month one year earlier and grows it, so seasonality survives instead of being averaged into twelve equal months. Presets by business type, services, subscription, retail, or a plain simple budget, give you a working model instead of an empty grid.
- Seed a year from your own actuals with a growth percentage
- Even spread or a per-period override on any line
- Business-type presets that start you with a working model
Budget versus actual, computed at read time
Budget vs actual is a query against the ledger for the month, or the year to date, so there is no refresh, no export and no stale copy circulating. Favorable and unfavorable resolve by account kind, a cost under plan reads as good, and it rolls up to a net-income variance.
- Live against the ledger for any month, or the year to date
- Favorable and unfavorable by account kind, net-income rollup
- Export the grid to CSV when someone insists on a spreadsheet
Forecasts that re-fit when the month closes
Closed months become actuals, open months get reprojected, and the reforecast tells you what changed and names the biggest movers. Projection runs three ways: a trend line with a seasonal factor per account, a bottoms-up pass that reads each customer and vendor separately, and won CRM deals feeding revenue.
- Trend plus seasonality per account, from a year of your own history
- Customer by customer, vendor by vendor: monthly, quarterly, annual, occasional or churned, each projected on its own cadence
- A reconciliation of what moved, with a written top-movers summary
Drivers you confirm, never just accept
The engine regresses each account against the signals it already has, revenue, every other line on the P&L, headcount, active customers and units sold, and proposes the relationship in words with the fit attached, for example that IT tracks headcount at about $5,100 a head on a $2,000 base with an R-squared of 0.94. It needs at least six aligned months before it will suggest anything, and nothing changes the forecast until a human confirms it.
- Driver inference by regression over your real history
- Ranked suggestions with the R-squared shown, and a revenue or account driver applies as a live formula
- Labor forecast built from the actual roster and approved hours
Cash by the week, and the runway under it
A direct cash forecast in weekly buckets, including the 13-week treasury view, or monthly if you plan that way. It starts from your book cash and adds the real events: open AR on its due dates, open AP on its due dates, scheduled debt service and contracted commitments not yet billed.
- Opening, inflow, outflow, net and closing on every bucket
- Flags the week you drop below your minimum cash
- Runway falls out of the same arithmetic, not a separate model
Variance explained, with the levers ranked
A number being off is not an explanation. Three decompositions run over the biggest movers: a margin bridge splitting a gross-profit miss into revenue effect and cost effect, a quantity-versus-rate bridge for driver-linked lines, and pipeline conversion for a revenue shortfall. Each pair sums exactly to the variance it explains.
- Margin, quantity and rate, and pipeline conversion bridges
- A plain-English cause written from the computed figures
- Recovery levers ranked and capped at what actually exists, so the pipeline lever never exceeds your open weighted pipeline
Budgets that reach the moment of spend
A department budget is not a separate store, it is the active budget version’s lines tagged to that department. Its actuals are the dimension-sliced income statement and its committed spend is the encumbrance from approved purchase orders, so the burn table shows budget, committed and actual side by side without double counting.
- Notified at 80% of budget, and again when it is crossed, as bills post against the department
- A funds check when a purchase order is approved, set to off, warn, or block the approval outright
- Financial targets with attainment tracked against real ledger figures
Pulse, zero-setup FP&A Early access
Pulse is a system-managed forecast version that builds itself from your ledger and refreshes on its own, alongside a business profile derived from the books: revenue concentration, cost behavior, how much of your revenue is recurring, and where the trend is going. You fork scenarios off it rather than editing it.
- A forecast and a proposed next-year budget with no setup project
- Scenario forks from a plain-language tweak, a growth rate, planned hires, a one-off purchase
- Available on request while it is in early access
Every number can tell you where it came from.
The reason a plan drifts from the books is that it lives somewhere else. This one does not, and the mechanics are worth being plain about.
Closed months are actuals
Not a copy of the ledger and not an import of it. When a period closes, the plan version stops guessing about that month and reads the posted figures.
Open months are projected, and the method is visible
A fitted trend times a seasonal factor, a counterparty’s own cadence, a driver you confirmed, or a number you typed. Each cell carries which one produced it, so a forecast is auditable in the same way an entry is.
Variance is arithmetic, not an export
Budget vs actual and the bridges under it are computed when you open them. The revenue effect and the cost effect sum to the gross-profit variance exactly, because that is an identity, not an estimate.
The plan never posts
A plan line is a plan line. Kate can draft a budget or a baseline forecast and a human applies it, and no forecast has ever written a journal entry.
Phase 1 is budgeting, forecasting, drivers, cash and variance, and we would rather name what is not built than let you find out later. The budget-vs-actual grid is company-wide today, so the dimensional reads are the department burn table and the variance explainer, which does scope to an entity or a dimension value. The balance sheet and cash flow are not yet driven off the plan version, so this is not an integrated three-statement model. There is no write-back to Excel or Google Sheets. There is no headcount-requisition, capital-expenditure or financing planning module, and no sensitivity or Monte Carlo modeling. Those are the next phases, not today.
“What’s our runway?” is a question, not a project.
Kate answers planning questions by running a real query against your ledger and your plan, never by doing arithmetic over text she retrieved. Ask her to build a budget and she drafts the version, you apply it. Chat and live voice both do it.
The planning tool stops being a separate bill.
The budgeting subscription, the cash-forecast spreadsheet and the variance deck someone rebuilds every month are one module here, sitting on the ledger the actuals come from. Nothing is exported between them because there is no between.
- Budgets and forecasting arrive on Launch, $99 a month, and carry through Growth and Scale. Solo at $59 is the ledger tier, so planning starts one step up.
- Planning by department, project or location needs Growth, $399, which is where dimensions land on every entry. That is also where purchase orders arrive, and with them the funds check on an approval.
- Multi-entity planning follows the ledger. The Multi-Entity & Global module is $600 a month flat, whether you run two entities or fifty.
- No per-seat fee, on any plan. The people who argue about the budget cost nothing to add, which is the only way a budget gets reviewed by the people spending it.
A plan is only as good as the ledger under it.
Bring last year. Leave with next year.
Seed a budget from your own actuals, re-fit the forecast against actuals at the first close, and see the variance explain itself instead of waiting for someone to write the deck.
Let’s begin