How the model thinks — and where it doesn't.
A feasibility tool is only as honest as the assumptions it makes visible. This page documents the engine, the benchmarks it draws on, and the limits any user should know before they act on a number we hand them.
One math engine, three depths
Quick Check (60 seconds), Full Check (5 minutes), and the full Pro Forma (free during private beta) all run on the same calculator. The deeper tiers don't change the math — they change how much of it we surface and how much input you give us to work with.
Industry-standard benchmarks
Prime cost healthy under 65%. Rent under 10% of revenue (NNN included). Loaded labor at 28-38% depending on service model. Buildout per square foot across light refresh, second generation, and ground-up custom. These are the same numbers a hospitality CPA uses.
Sanitize at the boundary
Every number that enters the engine is coerced and clamped before any arithmetic runs. We don't trust the form. Enums fall back to safe defaults. Negatives become zero where it makes sense. The calculator never throws on bad input — it just returns something defensible.
What the engine actually does
Revenue
- Dine-in covers × average ticket × seat-turn × days open. Takeout layered as covers × takeout average ticket × takeout share. Ghost kitchen revenue is orders × AOV with platform-vs-direct mix and platform commission applied separately.
- Per-business-type ramp curves: cafe 4 months to stabilize, restaurant 6 months, bar 9 months, ghost kitchen 3 months. The Year-1 monthly P&L applies these curves so the cash position chart isn't lying about month four.
- Seasonality vectors (12 months summing to 12.0) per business type, with concept-level overrides for lunch-forward concepts (sandwich, healthy bowls) and Q1-brutal concepts (full-service restaurants).
- CPG retail revenue is added separately when the operator has a parallel retail line.
Costs
- Food cost preset by concept style. Loaded labor cost by service model, with state-level tip-credit adjustments (CA/WA/MN no-tip-credit states +5pp, tip-credit states -3pp).
- Lease type with NNN load (NNN typically adds 25-40% on top of base rent). Occupancy cost includes utilities, R&M, insurance, and the NNN load — not just base rent.
- R&M reserve tiered by buildout level (1.0% of revenue for ground-up custom, 1.8% for light refresh of an old space).
- Pre-launch marketing budget × market-tier multiplier. Soft rent during buildout months. Pre-opening labor for training. Year-5 refresh capex tiered by equipment level.
Lender-grade math (under the hood)
- DSCR computed on EBITDA basis (the SBA-friendly convention) — drives the year-1 risk-flag for any deal with a real loan.
- Sources & Uses reconciled: equity injection + debt = startup costs + working capital reserve + ramp burn. Reconciliation surfaces in Section 10 as a single tie-out.
- Personal-guarantee implications are surfaced in the Risk Flags section, not buried in a banker appendix — this is the part every first-time owner needs to read before signing.
- Amortization on amortizing-loan basis. Depreciation on FF&E (7-year straight-line) and leasehold improvements (15-year straight-line) per IRS schedule.
- These are the lender-grade inputs to the dreamer-facing output — bankers can still ask for the underlying schedules, but the consumer-facing report doesn't lead with them.
The limits, in plain language
Not financial advice
This is an educational tool, not financial, legal, accounting, tax, or investment advice. Real-world results vary heavily by operator, market, lease, and execution. Before you sign a lease or write a check, talk to a CPA, a hospitality attorney, and at least one operator who has actually opened a place in your market.
Not a substitute for a real pro forma
We model what a careful, sober first-pass looks like. The Pro Forma tier produces a full operating model you can hand to a lender, but a hospitality CPA who knows your market, your build team, and your LOI will always produce a tighter forecast than any tool can.
Defaults are mid-market, not your market
We pre-fill defaults that produce realistic mid-market outcomes. Your specific lease, your specific menu, your specific GM — those move the math. Edit the inputs. Stress-test what you assume. Don't trust the headline number until you've challenged the assumptions behind it.
Investor returns are scenarios, not promises
The IRR / MOIC scenarios assume an exit at a multiple of year-7 net income. That's a useful framing for conversations with friends-and-family or LP investors. It is not a guarantee. Most independent restaurants are not bought; most operators take their return through distributions, not a sale.
Sources we draw on
The benchmark library is composed from public industry research, SBA underwriting guidance, hospitality CPA practice notes, and operator interviews. We don't claim originality on any single number — the value is in the synthesis and the per-concept tuning.
- Public industry surveys on prime cost, occupancy load, and ramp duration by concept type.
- SBA 7(a) loan underwriting guidance on DSCR thresholds, global cash flow analysis, and collateral coverage.
- Hospitality CPA practice notes on tip credit by state, buildout cost ranges, and depreciation schedules (FF&E 7-year, leaseholds 15-year).
- Operator interviews on real-world ramp curves, opening cash burn, and the “trough month” pattern that the cash position chart surfaces.
Kale Bittner — founder
Four years as COO and General Manager of Iron Triangle Brewing Company, scaling it from launch to 3,000+ barrels annually with distribution across Southern California. Built every forecasting and operational model from scratch in Excel, line by line, late at night — the way most first-time F&B operators still have to.
Then four years at Amazon Music / Wondery as Director of Marketing Strategy & Subscription Growth, building LTV models, subscription forecasts, and content-investment frameworks for a multi-million- subscriber business. Same modeling discipline, much better tools.
Restaurant Reality Check is the tool he wishes had existed when he was opening Iron Triangle. Read the longer story.
Run the check, then come back here.
Reading the methodology in the abstract is one thing. Reading it after the engine has just told you something uncomfortable about your concept is when it actually helps.