Clear assumptions.
Traceable results.

Model 2.2.2 uses 24-decimal fixed-point arithmetic. Currency inputs must be whole cents; nonzero sub-cent entries are rejected rather than silently rounded. Annual operating profit and coverage sum unrounded accrual profit before final display. Financial outputs are displayed to cents; loan schedules use cent-rounded payments and adjust the final payment to clear the balance. M1 is the entered revenue. Later revenue grows by the entered monthly rate. Variable costs scale with revenue using the baseline cost ratio. At zero baseline revenue, the entered variable-cost amount grows with the revenue-growth assumption and scenario multiplier; the ratio and break-even are undefined. Cash receipts, payments, taxes and debt are posted to cents at transaction boundaries; displayed net cash flow reconciles exactly to closing cash. Accrual revenue, expense and profit are rounded independently for display, so their displayed totals may differ from the unrounded annual result.

Operating expenses = variable costs + fixed costs + separate payroll + any ongoing plan cost. Fixed costs and separate payroll grow at the entered fixed-cost rate. Payroll is entered separately only if excluded from fixed costs. Cash taxes = max(0, operating profit) × the entered effective cash tax rate. This is a cash-planning assumption, not a tax calculation. No depreciation, tax losses, owner withdrawals or interest deductibility is modeled.

A month is modeled as 30 days. A 15-day collection delay collects half of a month’s revenue in that month and half in the next. A 30-day delay shifts receipts one month. Supplier delays apply to variable costs only; fixed costs and payroll are paid in the current month. Opening receivables and payables settle in M1. No earlier sales/purchases are assumed. Each invoice is posted to cents, its first delayed portion is rounded, and its residual is allocated to the following period; no cents are created or lost. Uncollected revenue and unpaid variable costs at M12 are shown as year-end receivables/payables. Their subsequent settlement is outside the horizon.

The baseline excludes the proposed loan and project. The plan adds the loan proceeds and deducts one-time investment in the selected start month. It then adds recurring revenue/cost impacts and the actual proposed loan payments for months within the horizon. Existing debt service is always deducted. The plan revenue assumption is subject to the same variable-cost ratio and optional capacity limit.

Lowest cash includes the opening balance and all 12 closing balances; ties use the earliest month. Funding gap = max(0, −lowest cash). Buffer shortfall = max(0, minimum cash buffer − lowest cash). Runway counts completed months before negative cash; an opening deficit means zero. “12+” only means no negative balance occurs in the forecast horizon.

Monthly rate r = annual nominal rate / 12. Level payment = principal × r / (1 − (1+r)^−term); at zero interest it is principal / term. The implementation uses the equivalent sum of discounted payments to avoid cancellation at very small positive rates. Each monthly schedule rounds interest and payment to cents and adjusts the final payment. Coverage = total operating profit / scheduled debt service actually due within the 12-month horizon. Plan coverage includes proposed payments beginning in the selected month. The lowest monthly coverage separately compares each month’s unrounded operating profit to that month’s debt service, so an annual surplus does not hide a monthly shortfall. Months without debt are excluded; no debt returns N/A. Coverage is an operating-profit proxy, not lender CFADS or a liquidity guarantee. This is a simplified illustrative estimate. Actual lender calculations and requirements vary by lender, product, borrower, and transaction.

Cash coverage uses posted receipts minus operating payments and modeled cash tax, divided by debt payments. Opening receivables/payables settle as modeled. Annual coverage uses all twelve months; monthly coverage excludes months with no debt. Shortfall months are detected from exact posted cents before ratio rounding. Opening cash, new loan proceeds and project capex are excluded from this indicator, but remain in the full cash forecast. A cash-coverage shortfall is not automatically a negative bank balance or a lender rejection. This is not lender CFADS or underwriting DSCR.

Operating break-even = (fixed costs + separate payroll) / contribution margin ratio. A non-positive margin has no attainable break-even. Targets above the supported display range are shown as unavailable, rather than an imprecise currency value. Baseline cash break-even = [fixed costs + separate payroll + existing debt / (1 − tax rate)] / contribution margin, assuming same-month receipts/payments and no growth. At full tax and positive debt it is unattainable. It excludes new financing, project spending and payment delays; with no debt it equals operating break-even. A supplied revenue capacity caps sales and triggers a warning when break-even exceeds it. Simplified pre-tax payback = max(0, project cost − financing) / incremental monthly cash benefit. Benefit uses the plan’s realized added revenue at the start month × contribution ratio, less added fixed cost and level loan payment. A non-positive or unknown benefit has no estimated recovery. A positive benefit that rounds to zero at currency precision also returns unavailable, with that reason; it does not establish zero economic benefit or permanent non-recovery. Zero initial equity does not remove loan obligations.

Unlevered incremental project cash = plan net cash + new loan repayment − loan proceeds − baseline net cash. It includes one-time project spending, operating changes, cash taxes and payment delays, while excluding financing transfers. Project NPV = sum of incremental cash in month t / (1 + annual nominal discount rate / 12)^t, for months one to twelve. All flows including project spending occur at month-end; the default annual discount assumption is ten percent and can be edited. No terminal value, post-horizon recovery, tax-loss carryforward, depreciation or interest tax shield is assumed. Cash recovery is the first month at or after project start when cumulative incremental cash is non-negative and remains so through M12. No initial outlay is shown separately; failure to recover within twelve months does not mean the project can never recover. The simplified equity payback above is a different pre-tax steady-state indicator.

Downside uses 80% of revenue, base 100%, and upside 120% across the full horizon. Growth then applies monthly. Variable costs adjust with sales; fixed costs do not change solely because of a scenario. These are stress tests, not forecasts of probability. Seasonality and capacity-expansion steps are not modeled.

The sample workspace uses fictional inputs; the planning workspace uses supplied assumptions. The policy registry contains dated official federal and selected state program references. The integrated Pennsylvania/California funding module screens the same business and project declarations used by the financial forecast and financing strategy, without estimating an award. The reviewed one-time project cash outlay is shared with the policy project budget; unreviewed amounts remain unknown. Its optional California search returns bounded official-portal leads, not verified eligibility; Pennsylvania uses a curated catalog and the official directory. Rules screen location, funding use, scale, hiring, training, technology and explicit application facts. Missing facts remain unknown. Catalogued closed intake overrides older open wording. A possible closure phrase detected on a retrieved page is only a review signal, not a confirmed program closure. Expired or incompletely reviewed sources require review. Current SBA ownership, residence, affiliate and underwriting requirements need lender or intermediary review; this screen does not collect or determine citizenship or residency. These are selected screening tracks, not complete legal eligibility tests. Agency and lender decisions remain separate.

One input snapshot, selected scenario and funding-plan choice feed the financial forecast, policy assessment and financing strategy. The strategy compares Base and Downside with that same plan choice, flags cash-buffer and debt-service pressure, and connects actionable program routes to unresolved conditions and next steps. Unawarded grants, incentives and published ceilings never enter forecast cash. Local JSON and PDF exports include all three modules; saved customer reports retain their original assessment and source evidence. Older saved reports do not receive invented historical strategy scores.

This platform-specific index is a transparent preparation checklist, not an official agency score, credit rating, probability of approval, or lender underwriting assessment. Each applicable check is worth twenty points. Ready checks receive twenty; unknown or attention-required checks receive zero; not-applicable checks are excluded from the denominator. The displayed index is the earned share of applicable points, rounded to a whole number. Criteria and reasons are listed beside the result. Unresolved input quality or program evidence makes the result provisional. A modeled coverage value of one only marks a cash arithmetic shortfall boundary; it is not an asserted lender approval threshold. Market demand, competition, legal compliance, collateral and credit history are not established by the calculation.

Financial calculations run before AI explanations. If enabled by the owner, the server recalculates submitted inputs, removes the business name, and sends structured numerical facts to OpenAI for a draft and Anthropic for an independent review. AI prose is separate from authoritative figures, constrained to the supported English character set, schema-checked, and rejected if it introduces numeric figures. Unsupported scripts and hidden control characters are rejected before delivery. Model output still requires human review. The configured models are GPT-6.1 Sol and Claude Sonnet 5.5. Provider requests require enabled service, credentials, shared budget configuration and explicit consent. Customer reports additionally require an authorized account and allowance or credit; the separate owner preview uses an access code. Model review is not a promised human review. Policy explanations use only the server-selected rule results and citations. Official pages are retrieved before a paid policy explanation. Candidates with inaccessible sources or unresolved availability warnings are omitted and marked; if all selected candidates lack retrievable sources, the call stops before charging. The independent reviewer cannot add programs, citations or qualification claims. Retrieval does not silently update reviewed rules. Saved reports retain their source evidence and policy assessment; reopening them does not substitute the current catalog for their historical citations.

Free calculation drafts stay in page memory and are cleared when you leave or refresh. Downloads are local. The sample workspace requires fictional data. Enabled customer accounts store account and usage records; explicitly requested AI reports send structured assumptions and calculated facts to OpenAI and Anthropic, excluding the business name. Name removal does not guarantee anonymity. Saved reports expire after thirty days unless deleted earlier; minimal receipts are retained separately. Official searches send structured sector/purpose terms and applicant filters, not the business name or financial figures. Hosting and external services may process technical logs. Read the linked privacy and legal notices for details.