Know your numbers.
Then know what they mean.

Where it breaks. Comparing ARR growth to revenue growth and calling it the same number. Pick one, label it, use it consistently in every document.

Worth knowing. Private B2B SaaS median: 22% (SaaS Capital, CY-2025). Bootstrapped 20%, equity-backed 25%.

Where it breaks. CAGR hides a bad most-recent year. If last year was flat, the buyer will find it — show the year-by-year alongside.

Worth knowing. Use for 3+ year histories. Below 3 years, show the actual years — CAGR on two points is theater.

Where it breaks. The accrual trap from the deck lives here. A 12-month prepay is 1/12 of that cash per month — booking it as one month of revenue inflates ARR and is caught in the first hour of diligence.

Worth knowing. Count only committed, recurring contracts. Exclude one-time fees, services, and usage overage above committed minimums.

Where it breaks. Netting expansion against churn. They are different diagnoses — expansion masking heavy churn is a leaky bucket priced as a growth story.

Worth knowing. Four components, always. If you can only produce net growth, that itself is the finding.

Where it breaks. Including new logos in the cohort. The cohort is frozen at day one; only what happened to those customers counts.

Worth knowing. Bootstrapped $3–20M ARR median 103% (SaaS Capital 2026). Self-serve/PLG runs far lower — ChartMogul median 82%.

Where it breaks. Leading with NRR and never showing GRR. A sophisticated investor asks for GRR specifically because it's the number you can't dress up.

Worth knowing. Median 84%, down from 88% (Benchmarkit/Aleph, CY-2025). Enterprise 91%, sub-$5K ACV 80%.

Where it breaks. Measuring monthly and multiplying by 12. Churn compounds — annualize as 1 − (1 − monthly)^12.

Worth knowing. Logo churn runs higher than revenue churn — small accounts leave first. Enterprise implies ≤9%/yr; sub-$5K ACV closer to 20%.

Where it breaks. Thinking this only costs you multiple. It more often costs you deal structure — 30–50% of price pushed into a 24-month earnout tied to that customer staying.

Worth knowing. Above 20%, buyers discount the concentrated slice of EBITDA separately at a lower multiple and blend. Above 30%, expect earnout and escrow, not just a lower price.

Where it breaks. Calling repeat customers 'recurring.' Recurring means contracted. A customer who happens to reorder is repeat revenue — a real asset, but it prices differently.

Worth knowing. Contracted and recurring is worth multiples of project or one-time revenue at the same dollar amount.

Where it breaks. Treating a healthy cash balance as profit when it is over-billing — customer money for work you have not performed. Contractors spend it and hit the back half of the job with no funding. The mirror error is never updating estimated cost at completion, which hides job fade until it is unrecoverable.

Worth knowing. GAAP percentage-of-completion applied operationally. Surety underwriters conventionally treat aggregate under-billings above about 1% of annual revenue, or any single job under-billed by more than 10% of contract value, as a red flag. Underwriting convention, not published data.

Where it breaks. Hosting, direct labor, support, and delivery software belong in COGS — not OpEx. Moving them up inflates gross margin and is the first thing a diligence team re-maps.

Worth knowing. SaaS software median 80%, blended 76% (CY-2025). Services ~30%. Retail/ecommerce ~33–35%. Usage-based pricing 62%.

Where it breaks. Treating gross margin and contribution margin as synonyms. Contribution margin also strips variable selling costs — commissions, payment processing, shipping.

Worth knowing. Run it per product, per channel, per customer segment. The blended number hides the loser.

Where it breaks. Adding back depreciation in a capex-heavy business as if it were free. If equipment genuinely must be replaced, sophisticated buyers deduct maintenance capex right back out.

Worth knowing. SBA underwriting starts from EBITDA, then deducts cash taxes, owner distributions and unfinanced capex to reach cash flow available for debt service. Your debt capacity is built on that lower figure, not on raw EBITDA.

Where it breaks. Building an add-back schedule you can't source. A quality-of-earnings review tests each line — undocumented add-backs don't get negotiated down, they get removed.

Worth knowing. Every add-back needs a document behind it. Standard: one-time legal, flood/disaster, above-market related-party rent. Contested: travel, meals, family payroll, club dues.

Where it breaks. The most expensive mistake on this sheet — see #45. SDE and EBITDA multiples are not interchangeable.

Worth knowing. Convention switches on enterprise value, not revenue: under ~$2M EV use SDE; $2–3M EV is the transition zone where both appear; $3M+ EV use EBITDA. Brokers follow this split in their own reporting.

Where it breaks. Presenting net margin to a lender who is underwriting to EBITDA, or to an investor who is underwriting to operating leverage. Know which audience wants which line.

Worth knowing. Operating margin is the cleaner comparison across companies; net margin is distorted by capital structure and tax posture.

Where it breaks. Using gross margin instead of contribution margin, which understates the break-even point and makes the plan look safer than it is.

Worth knowing. Recalculate after every material hire or lease. It moves more than people expect.

Where it breaks. Payroll landing in one undifferentiated bucket. Until people are mapped to departments, no forecast built on this is defensible.

Worth knowing. SaaS medians (CY-2024): S&M 37%, R&D 34%, G&A 24%. VC-backed run S&M ~47%; PE-backed ~33%. These are separate medians from a VC-weighted panel — they are not a budget, and they do not sum to a viable cost structure for a bootstrapped company.

Where it breaks. Starving real infrastructure to move this ratio. Underpaying finance and technology to look efficient is the overhead myth, and it shows up two years later as a failed audit or a lost grant. The ratio measures allocation, not effectiveness.

Worth knowing. BBB Wise Giving Alliance Standard 8 requires at least 65% of total expenses on program activities. Charity Navigator's finance score rewards materially higher — around 75% and up. Both read the figure off the Form 990 functional expense statement, so the allocation you book is the allocation you are judged on.

Watch. Whether your allocation method between program, management and fundraising is documented and applied consistently. An auditor will ask.

Where it breaks. Excluding salaries and loaded comp — counting only ad spend. Fully-loaded CAC includes the people, their benefits, and the tools.

Worth knowing. Median new-customer CAC ratio: $2.00 of S&M per $1 of new ARR (software). There is no cross-industry CAC benchmark — a home-services business and a software business have nothing comparable here, so track your own trend and compare only within your industry.

Where it breaks. Omitting gross margin. Paying back on revenue rather than gross profit understates payback by roughly a quarter.

Worth knowing. Median 16 months, top quartile ≤6 (Aleph/Benchmarkit, CY-2025). SMB ACV ~11 months; enterprise ~22. Both terms must be per-customer or both aggregate — mixing them yields a meaningless number. Tiers are at-scale norms; read them against your ACV segment, not instead of it.

Where it breaks. Using revenue instead of gross profit, and a churn rate borrowed from a benchmark rather than your own cohorts. Both inflate LTV, and they compound.

Worth knowing. ARPA and churn must be on the same period — monthly with monthly, annual with annual. Use revenue churn, not logo churn. Only meaningful with real churn history; below ~18 months it is a projection, not a measurement. If NRR exceeds 100% the formula breaks down entirely.

Where it breaks. Leading with LTV:CAC in a pitch. Sophisticated investors discount it precisely because both inputs are assumption-heavy — lead with CAC payback instead.

Worth knowing. A rule of thumb, not benchmark data — the 3:1 convention comes from a survey panel. Above 5× may signal underinvestment in growth.

Where it breaks. Running it on a single quarter. It's noisy — use a trailing four-quarter view before drawing a conclusion.

Worth knowing. Median 1.37 (CY-2025), up from 0.94 — first time above 1.0 in four years. 75th percentile 2.14.

Where it breaks. Using gross burn. The multiple is defined on net burn — gross overstates it and makes an efficient company look wasteful.

Worth knowing. Tiers are David Sacks / Craft Ventures: <1× amazing, 1–1.5× great, 1.5–2× good, 2–3× suspect, >3× bad. These are at-scale norms — early stage naturally runs higher, ~3.4× at $0–1M ARR falling to ~1.4× by $25–50M. Read the tiers against your stage, not instead of it.

Where it breaks. Mixing an ARR growth rate with a GAAP EBITDA margin. Both terms must come from the same basis, or the score is meaningless.

Worth knowing. Median 25% (CY-2025) — more than half the market is below the line. Top quartile 43%+. Only becomes meaningful around $20M ARR; below that it's volatile.

Where it breaks. Counting contractors inconsistently between periods, which manufactures a trend that isn't there.

Worth knowing. Private SaaS median $141K ARR/FTE (SaaS Capital 2026); $1–3M ARR ~$110K, bootstrapped higher at every band. Those figures are software-only. For a business that resells materials or subcontracts, revenue per employee flatters badly — use Labor Efficiency Ratio instead, which measures gross profit per labor dollar.

Where it breaks. Forecasting headcount at base salary. Across ten hires that understates burn by roughly three extra salaries.

Worth knowing. Typically 1.25–1.4× base salary. Run it before the offer, not after the first payroll.

Where it breaks. Judging a new donor-acquisition program on first-year cost. Acquisition is bought at a loss and repaid over the donor's lifetime; measuring year one alone kills the programs that would have funded you in year five.

Worth knowing. BBB Wise Giving Alliance Standard 9 sets the ceiling at 35 cents per dollar of related contributions. Channels differ enormously underneath any blended figure — a major-gift program and a cold direct-mail acquisition campaign are not comparable, and averaging them hides both.

Watch. Cost per channel, not blended. And whether the denominator includes the grants you would have received anyway.

Where it breaks. Quoting gross when asked for net, or averaging a quarter that included an annual prepay. Show the monthly series.

Worth knowing. Compute from the bank, not the P&L. Burn is a cash concept.

Where it breaks. Ignoring committed but unspent obligations — signed leases, annual software renewals, severance. Real runway is shorter than the arithmetic.

Worth knowing. Use a trailing 3-month average burn, and state whether the figure is before or after any planned raise. If you are cash-flow positive, report it as not applicable rather than as a negative number.

Where it breaks. Presenting EBITDA as if it were cash. In a working-capital-hungry or capex-heavy business the two diverge sharply.

Worth knowing. The gap between EBITDA and FCF is where working capital and capex intensity show up. Buyers look at both.

Where it breaks. Averaging DSO across a book with one very large slow payer. Look at the aging, not just the average.

Worth knowing. Small business target 15–30 days. Cross-industry median ~36. The AR triage red flag — items over 60 days — lives here.

Where it breaks. Stretching DPO right before a sale. The working capital peg (#46) is built on your normal behavior, so you hand the benefit straight back at close.

Worth knowing. Read alongside DSO. Paying in 20 while collecting in 50 is a self-inflicted cash squeeze.

Where it breaks. Carrying obsolete inventory at cost. It flatters turns and the balance sheet, and gets written down in diligence.

Worth knowing. Highly industry-specific. Track your own trend before comparing to anyone else's.

Where it breaks. Benchmarking against public-company data. SMB working capital is structurally worse — weaker collections and less supplier leverage. Use it for shape, not as a target.

Worth knowing. Broad US baseline ~32 days. Software runs near zero or negative; distribution 35–65; manufacturing 50–90; construction 60–120.

Where it breaks. Counting undrawn credit line availability as a current asset. It isn't one, and the drawn balance is a current liability.

Worth knowing. Lenders treat 1.0 as the floor; 1.5+ reads as comfortable.

Where it breaks. Including receivables over 90 days. If they're not collectible this quarter they don't belong in a liquidity test.

Worth knowing. Lenders want ≥1.0. The gap between your current and quick ratio is the size of your inventory bet.

Where it breaks. Using average inventory at retail rather than at cost in the denominator. That deflates GMROI by roughly your margin percentage and makes every category look like a failure. Second: a single point-in-time inventory figure instead of a twelve-month average, which wrecks the number for any seasonal retailer.

Worth knowing. $2.00 is the common healthy floor but it is heavily category-dependent. Retail Owners Institute five-year averages: apparel $1.50–$3.00 · specialty $3.00–$15.00 · furniture $2.00–$4.42 · consumer electronics $5.00+ (appliances $2.30) · health and beauty $2.85–$6.67 · motor vehicles $0.59–$3.19. Directional — the underlying data is undated.

Where it breaks. Counting only warehouse rent. The capital cost — what that money would otherwise earn, or what your line of credit charges — is usually the largest single component and is almost always omitted, which makes over-stocking look free. At a 25% carrying cost, $200K of excess inventory costs you $50K a year.

Worth knowing. Carrying costs run 20–30% of average inventory value per year, attributed to the Institute for Supply Management. Retail shrink runs 1.4–1.6% of sales per the NRF/LPRC National Retail Security Survey — note the most-cited figure is 2022 data published in 2023.

Where it breaks. Counting restricted net assets as reserve. Money a donor designated for a purpose is not available to make payroll, and treating it as though it were is the most common way a non-profit discovers it is illiquid.

Worth knowing. Three to six months is the range most often cited, and many funders ask for it directly. The National Council of Nonprofits is explicit that a single universal standard is a myth — the right level depends on how volatile and how restricted your revenue is. A grant-reimbursement organization needs more than one with recurring unrestricted giving.

Watch. Whether the board has adopted a written reserve policy naming a target and the conditions for drawing on it. Without one, the reserve gets spent by default rather than by decision.

Where it breaks. Using net income instead of EBITDA, or forgetting that the new debt is included in the denominator. You're proving you can cover the loan you don't have yet.

Worth knowing. Per SOP 50 10 8 (Procedural Notice 5000-875701, eff. Mar 1, 2026): 7(a) Small Loans — now defined as ≤$350K, lowered from $500K — require 1.10:1 historical and/or projected. Change of ownership 1.15:1 through Sep 30, 2026. Conventional banks typically want 1.20–1.25×. SBA 504 SOP floor is 1.1:1; CDCs underwrite to 1.15–1.25×. Note the bank starts at EBITDA then deducts cash taxes, owner distributions and unfinanced capex — your DSCR will come out lower than raw EBITDA implies.

Watch. SOP 50 10 8.1 takes effect Oct 1, 2026: acquisition and owner-buyout loans move to 1.25×, and post-closing projections are no longer permitted — coverage must come from the last fiscal year end or a two-year average. A deal financeable on or before Sep 30, 2026 may not be financeable after it. The trigger is the SBA loan number issuance date, not the application date.

Where it breaks. Preparing the business file and ignoring the personal one. A clean company with a stretched personal balance sheet still gets declined.

Worth knowing. SBA baseline 1.15×; most lenders want 1.25×+. Income side includes owner and spouse W-2, net rental, dividends. Debt side includes personal mortgage, autos, student loans, card minimums, support obligations — verified against credit reports and three years of personal returns.

Watch. Owner-compensation add-backs now face a global cash flow test — the salary you claim you can live on must reconcile to your documented personal expenses. Aggressive comp add-backs now fail twice.

Where it breaks. Reading the number and not the definition. Owner distributions are often pulled into 'fixed charges' by the credit agreement — check the contract language, not the ratio.

Worth knowing. Typical covenant 1.0–1.25×. Definitions vary by credit agreement — leases, preferred dividends and insurance premiums are often pulled in. Whatever sits in the denominator must be added back in the numerator, or you penalize the same dollar twice.

Where it breaks. Relying on it for amortizing term debt, where DSCR and FCCR dominate because they include principal.

Worth knowing. Covenant floors commonly set at 2.0×. Below 1.5× reads as distress.

Where it breaks. Excluding capital leases and seller notes. Lenders include them; your internal number should too.

Worth knowing. Below 3.0× reads as healthy for SMB bank lending. Above ~3.5×, conventional lenders generally want sponsor equity or subordinated debt. This is strong convention rather than a published rule.

Where it breaks. A negative equity balance from years of distributions exceeding earnings. Common, fixable, and much better explained by you than discovered by them.

Worth knowing. Above 2.0 reads as higher risk in SBA lender guidance, though tolerated at startup and expansion stage. Debt-to-net-worth below 1.0 reads as stable. With negative equity the ratio is meaningless — report the equity deficit itself, not the ratio.

Where it breaks. Borrowing a multiple from the wrong dataset. PE-sponsored deals ≥$10M run 6.4×+ — applying that to a $2M-EBITDA business is the classic and most expensive error on this sheet.

Worth knowing. Broker-intermediated deals (IBBA Q1 2026): <$500K → 2.0× SDE · $500K–1M → 2.8× SDE · $1–2M → 3.0× SDE · $2–5M → 4.0× EBITDA · $5–50M → 4.5× EBITDA. BizBuySell Q1 2026, 2,345 closed deals: 2.7× SDE, median price $350K.

Where it breaks. Planning your life around enterprise value. A '$30M deal' routinely becomes $20M cash at close, $5M earnout, $3M escrow, $2M seller note.

Worth knowing. Deals are structured cash-free, debt-free. Then subtract escrow, holdback, earnout, and seller note to get cash at close.

Where it breaks. Reaching for a revenue multiple because the EBITDA multiple is unflattering. Buyers notice, and it signals the earnings won't survive scrutiny.

Worth knowing. SMB five-year average ~0.67× revenue. Used seriously only for high-growth recurring-revenue businesses.

Where it breaks. Anchoring on the larger multiple. Ask any advisor quoting a multiple: 'of SDE, or of EBITDA?'

Worth knowing. Worked example: $1M SDE with $250K owner comp = $750K EBITDA. 3× SDE = $3M. 4× EBITDA = $3M. Same price. The EBITDA multiple only looks higher because the base is smaller.

Watch. Applying an EBITDA multiple you read online to your SDE overvalues the business by roughly a third. Always confirm which base a quoted multiple refers to before you anchor on a number.

Where it breaks. Sprinting to collect AR and stretch AP before close. The peg is built from your own historicals, so you deliver below a normal-behavior target and pay the difference back — the adjustment that surprises sellers most.

Worth knowing. Usually set as a trailing 12-month average of monthly NWC (~70% of deals). Collars of ±5–10% create a no-adjustment zone. Final true-up lands 60–90 days after close.

Where it breaks. Computing ownership on issued shares rather than fully-diluted. Options, warrants, and SAFEs all convert.

Worth knowing. If the pool is created pre-money, the existing shareholders fund it alone — an effective additional dilution the headline valuation hides.

Where it breaks. Assuming a headline valuation translates proportionally to founder proceeds. With a preference stack, a mid-range exit can return far less to common than the ownership percentage implies.

Worth knowing. Run the waterfall at several exit values before signing, not after receiving an offer. Participating preferred and stacked preferences change the answer sharply at lower exit prices.

Where it breaks. Revising the budget to match the actuals. That erases the only signal the exercise produces.

Worth knowing. Run it monthly by department. The pattern of variance matters more than any single month.

Where it breaks. Never scoring yourself. Investors who fund you will build this record whether or not you do.

Worth knowing. Keep a rolling record of prior forecasts versus outcomes. A modest projection you consistently hit outweighs an ambitious one you consistently miss. Bounded above at 100% but not below — a forecast double the actual scores 0%, triple scores −100%, and undefined when actual is zero. Use it on revenue lines, not on items that can hit zero.

Where it breaks. Modeling monthly. Payroll on the 15th against a receivable landing on the 28th is invisible at monthly resolution and fatal in practice. The second mistake is modeling an average week — real outflows are lumpy, and the low point is what matters, not the closing balance.

Worth knowing. Built from AR aging, AP aging, payroll dates and debt service. Roll it forward every week. The point of weekly resolution is that payroll and collections land on different days: the average week can look fine while a specific week does not.

Where it breaks. Meeting your first QoE as the seller. Running a sell-side quality of earnings before going to market means you find the adjustments while you can still fix them.

Worth knowing. Under SOP 50 10 8.1 (eff. Oct 1, 2026), mandatory on SBA acquisition loans at $3M+ purchase price, and it must be lender-commissioned — a buyer's own QoE does not satisfy the requirement. Confirm current requirements with your lender before relying on this.

Where it breaks. Adding a 30% markup and believing you earned a 30% margin. You earned 23.1%. On $1M of revenue that is a $69,000 hole you will look for in the wrong place all year.

Worth knowing. Target-margin → markup: 30% margin needs a 1.43× multiplier · 40% needs 1.67× · 50% needs 2.00× (keystone). Convention, not survey data — the arithmetic is exact.

Where it breaks. Testing the discount against gross margin instead of contribution margin — it makes the required volume lift look smaller than it is. And assuming the discount only reaches new buyers; it reaches everyone who would have paid full price.

Worth knowing. At a 30% contribution margin a 10% discount requires roughly 50% more volume just to stand still. At 40% CM a 10% price rise lets you lose 20% of volume and come out level. Method per Nagle & Müller, The Strategy and Tactics of Pricing.

Where it breaks. Leaving payroll taxes, workers' comp and benefits out of the labor half. That alone understates prime cost by 3–6 points and moves an operator from 'problem' to 'fine' on paper. Second: running it monthly. Monthly prime cost is a post-mortem — it has to be weekly.

Worth knowing. Target 60–65% of sales; limited-service should land below full-service. National Restaurant Association 2025 Operations Data Abstract (Aug 2025, 900+ operators, 2024 data): full-service labor 36.5% median — 34.2% for profitable operators, 42.9% for loss-making. Limited-service 31.7% median. That 6.1-point labor gap between profitable and unprofitable is the number to hold onto.

Where it breaks. Expressing variance as a percentage of COGS instead of sales — that inflates it roughly threefold and makes the published bands useless. Second: counting inventory on a different day each period, which manufactures variance that isn't there.

Worth knowing. Under 2% is well controlled · 2–3% acceptable · 3–5% needs work · above 5% investigate now. Practitioner convention, consistent across restaurant-tech sources; no published dataset.

Where it breaks. Ranking by margin percentage instead of contribution dollars. A $4 coffee at 85% contributes $3.40; a $28 entrée at 65% contributes $18.20. Percentage-ranking tells you to push the coffee and cut the entrée, which is how a business engineers itself into insolvency.

Worth knowing. Kasavana & Smith menu-engineering convention (1982), still the practitioner default: an item is 'popular' above (1 ÷ number of items) × 0.70 of total units. Star = high margin and high volume · Plowhorse = low margin, high volume · Puzzle = high margin, low volume · Dog = neither.

Where it breaks. Using revenue instead of gross profit. Revenue per employee flatters any business that passes through materials or subcontractors — a contractor at $3M revenue with $1.8M of materials looks efficient and is drowning. Second: leaving the owner's labor out. If the owner works sixty hours and takes distributions, this number is fiction.

Worth knowing. Greg Crabtree (Simple Numbers, Straight Talk, Big Profits, 2011) popularized it and deliberately declines to publish a universal target; he benchmarks against your own rolling 12-month trend. The repeated working targets of 2.0× direct and 4.0× management are practitioner convention with no dataset behind them — treat them as a starting point, not a standard.

Where it breaks. Dividing by 2,080 hours. The denominator has to be productive hours — take out PTO, holidays, training, travel and shop time. A tech paid for 2,080 who bills 1,400 costs roughly 48% more per billable hour than the 2,080 figure suggests. This one error underprices more trade work than anything else on this page.

Worth knowing. BLS Employer Costs for Employee Compensation, March 2026 (released June 12, 2026): private-industry total compensation $46.60/hr, wages $32.60 (69.9%), benefits $14.01 — a 30.1% load, or a 1.43× multiplier. The familiar 1.25–1.40× rule of thumb is low against current data. Construction runs 35–60% depending on trade; employer FICA is 7.65% to the $184,500 Social Security wage base (SSA, 2026).

Where it breaks. Recovering overhead across available hours rather than historically achieved billable hours. Spread overhead over 2,080 and sell 1,400, and you under-recover a third of your overhead every year while wondering why a 30%-gross-margin business has no profit.

Worth knowing. Method is practitioner convention. ServiceTitan's worked example: 5 techs × 583 billable hours = 2,916 hours; $81,648 burdened labor + $100,000 overhead ÷ 2,916 = $62.29 break-even per sold hour. Note that implies about 28% billable utilization against 2,080 available hours — realistic for residential service, and a useful shock.

Where it breaks. Measuring utilization and never realization. A shop at 80% utilization and 70% realization is worse off than one at 70% and 95%. Owners chase the visible number and bleed on the invisible one — scope creep, write-offs and courtesy discounts.

Worth knowing. Billable utilization fell to 66.4% in 2025, the lowest in the survey's history, against an optimal threshold near 75%; project margin 37.7%; revenue per billable consultant $210K, up 6% year over year. SPI / Deltek 2026 Professional Services Maturity Benchmark.

Where it breaks. Averaging across the whole week. A restaurant at 45% average seat utilization may be at 95% Friday and Saturday and 20% Monday to Wednesday. The average says add capacity; the daypart says change your Tuesday pricing. Run this by day or daypart or it will actively mislead you.

Worth knowing. Capacity units are seat-hours, chair-hours, room-nights, bay-hours, class slots or truck-days. RevPASH originates in Sheryl Kimes' Cornell revenue-management work; the generalization to gyms, salons, clinics and trades is convention. Restaurant table-turn convention: 1.5–2 turns for casual dining at dinner, 3+ for fast casual.

Where it breaks. Comparing base rent only. CAM, triple-net taxes and insurance routinely add 25–40% on top, and percentage-rent clauses add more as you grow. Owners sign to a 7% base rent and operate at a 10% all-in occupancy cost.

Worth knowing. Retail convention targets base rent at 5–10% of gross annual sales, with the full range across business types running 2–20%; restaurants 6–10%, salons 8–12%, professional services tolerate up to ~15%. These are broker convention, not survey data — the commercial-real-estate sources that publish the formula decline to publish healthy ranges.

Where it breaks. Counting only the job ad and the recruiter fee — usually under a fifth of the real cost. The dominant components are vacancy coverage (overtime at 1.5×, or revenue simply not earned) and the ramp period. A restaurant replacing a line cook counts a $200 ad and ignores six weeks of a slower kitchen.

Worth knowing. The familiar '50–200% of salary' figure is repeated everywhere and traces back to a 2012 Center for American Progress study; it is not a current, primary benchmark. This calculator builds the number from your own components instead, which is both more defensible and more useful. Use the 50–200% range only as a sanity check on the result.

Where it breaks. Using the base hourly rate as the FLSA 'regular rate'. If you pay a nondiscretionary bonus or a shift differential and compute overtime on base pay alone, you have underpaid — the most common wage-and-hour class-action trigger, and back pay is doubled as liquidated damages. Separately: comparing a few overtime hours against a full-time salary and concluding overtime always wins. Compare per hour, then ask whether you can fill the rest of the year.

Worth knowing. Overtime carries only variable burden — FICA, workers' comp, SUTA under the state base. Health premiums and most fixed benefits do not scale with hours, which is why overtime is usually cheaper per hour than it looks. The real question is rarely the hourly rate: it is whether you have enough sustained work to fill a new hire's whole year.

Watch. The 2026 'no tax on overtime' deduction ($12,500 single / $25,000 MFJ, phasing out from $150K/$300K MAGI, tax years 2025–2028) is an employee income-tax deduction on the premium half only. It does not reduce employer FICA, FUTA, SUTA or workers' comp — overtime did not get cheaper for the business, and employers picked up a new reporting burden.

Where it breaks. Comparing a stated rate to an APR, and assuming a '10% loan' costs 10% of principal in total interest. On a fully amortizing loan the average balance is about half the principal, so a five-year 10% loan costs roughly 27% of principal in total — not 50%. Owners optimize the monthly payment and buy more total interest.

Worth knowing. Bank prime was 6.75% at September 2, 2026 (Federal Reserve H.15). SBA 7(a) Working Capital Pilot caps run base rate +6.5% on loans ≤$50K down to +3.0% above $350K — roughly a 9.75–13.25% band at that prime, which is the right yardstick for anything else you are offered.

Where it breaks. Reading '1.3 factor' as '30% interest'. It is about 30% of face over a period in which you hold, on average, roughly half the money. Even the standard cost ÷ advance × 365 ÷ days formula that most calculators use materially understates the APR, because it treats the whole principal as outstanding for the whole term while daily remittances retire it continuously.

Worth knowing. Factor rates run 1.1 to 1.5. The Fed's 2026 Small Business Credit Survey (published March 3, 2026) found 60% of firms borrowing from online lenders said actual costs were higher than expected, against 37% at small banks and 32% at large banks — online fintech share of applicants rose from 17% in 2020 to 29% in 2025.

Watch. Shorter terms make an advance MORE expensive, not less — the opposite of every intuition about loans. The same 1.35 factor is roughly 70% APR over twelve months and roughly 139% over six. If you are offered a renewal, check whether the payoff of the existing advance includes the unearned portion of the original factor: rolling one advance into the next charges a factor on money you have already paid for.

Where it breaks. Two. 'It's only 2%' — it is 37% annualized. And paying on day eighteen: you have paid the full price and captured none of the float. Either pay on day ten or ride to day thirty; there is no partial credit.

Worth knowing. 2/10 net 30 is 37.2% annualized (44.6% compounded). Others: 1/10 net 30 = 18.4% · 2/10 net 45 = 21.3% · 3/10 net 30 = 56.4%. With bank prime at 6.75% and a line of credit near 10%, borrowing to take 2/10 net 30 is almost always correct.

Where it breaks. Comparing headline rates instead of computing the effective rate off an actual statement. Flat-rate pricing is dominated by the fixed per-transaction fee at small ticket sizes: at 2.6% + $0.10, a $10 ticket costs 3.6% and a $200 ticket costs 2.65%. A coffee shop and a contractor on the same processor pay very different rates.

Worth knowing. Interchange typically runs 1.15–3.15% depending on card type and how the transaction is entered; all-in effective rates for small merchants usually land between 2.2% and 3.5%. Rewards and corporate cards sit at the top of the range, so B2B merchants run structurally higher.

Where it breaks. Comparing the monthly lease payment to the monthly loan payment. That ignores the residual — you own an asset at the end of one and nothing at the end of the other. And a $1-buyout lease is not a lease: for tax purposes it is a conditional sale, so you depreciate the asset and deduct interest, not the payments.

Worth knowing. 2026 tilts hard toward buying for a profitable business: 100% bonus depreciation is permanent for property acquired after January 19, 2025 under OBBBA, and §179 allows $2,560,000 with phase-out beginning at $4,090,000 (Rev. Proc. 2025-32). §179 is limited to business income and carries forward; bonus depreciation is not so limited and can create an NOL instead.

Where it breaks. Applying the percentage to revenue instead of net profit. After that: missing the 110% safe-harbor step-up when prior-year AGI was over $150,000, which is the single most common penalty cause for a growing business, and forgetting state estimates entirely.

Worth knowing. Federal only, 2026, pass-through owner. Uses the verified constants: SE 15.3% on 92.35% of net profit, Social Security wage base $184,500 (SSA), standard deduction $16,100 single / $32,200 MFJ and the IR-2025-103 brackets, §199A threshold $201,750 / $403,500 (Rev. Proc. 2025-32). Safe harbor is 90% of the current year or 100% of the prior year — 110% if prior-year AGI exceeded $150,000. The familiar 'save 25–30%' rule is federal-only and breaks in any income-tax state. Run at these constants, a pass-through owner with $100K–$250K of profit and no other income typically lands between 24% and 29% federal, but the answer moves several points with filing status, spouse income and whether §199A survives.

Watch. The four estimate periods are not equal in length — April 15 covers three months, June 15 two, September 15 three and January 15 four. A business that earns most of its money in Q4 and pays four equal installments still gets penalized for the earlier quarters. If your income is lumpy, ask your preparer about the annualized income installment method on Form 2210 Schedule AI.

Where it breaks. Quoting the bracket and forgetting SE tax entirely, then using that wrong number to decide on equipment, retirement contributions and whether a spend is 'worth it after tax'.

Worth knowing. The next dollar of Schedule C profit gets hit by SE tax, then half of that SE tax is deducted, then QBI is reduced by the same deduction, then 20% of QBI comes off. Federal only, 2026. Below the $184,500 Social Security wage base a 22%-bracket owner is deciding at 30.5%; a 24%-bracket owner at 32.0%.

Watch. Your marginal rate FALLS when you cross $184,500 of net self-employment earnings, because the 12.4% Social Security component switches off. If your income is lumpy, that is an argument for pushing income into a year you are already over the base.

Where it breaks. A zero or nominal salary on substantial profit — the exact fact pattern the IRS reclassifies. And forgetting that lower W-2 wages shrink Solo 401(k)/SEP contribution room and reduce future Social Security benefits.

Worth knowing. There is no statutory or IRS basis for the '60/40 rule' — the standard is comparable-market compensation for the services actually performed. IRS guidance (S Corporation Compensation and Medical Insurance Issues, updated March 3, 2026) lists nine factors and cites Watson v. United States. Incremental cost typically runs $1,500–$4,500 a year for payroll service, the 1120-S and state entity taxes; California adds an $800 minimum franchise tax plus a 1.5% S-corp tax.

Watch. Above the §199A threshold ($201,750 single / $403,500 MFJ) the conventional advice inverts. The non-SSTB QBI limit is 50% of W-2 wages, so cutting salary destroys deduction: at a 32% bracket you lose 16¢ of QBI per dollar of salary cut, against 2.9¢ of Medicare saved once you are past the wage base. For a high-income non-SSTB owner, the right move is to RAISE salary, not cut it.