Dog #2 says: "RevPAR is the heartbeat of a hotel model. If RevPAR is growing, everything else can be fixed. If RevPAR is declining, no cost cut saves you."

Hotel models center on rooms as the fixed-cost base and revenue per available room (RevPAR) as the top-line driver. Once you've paid for the building, the marginal cost of filling a room is tiny — so occupancy rate swings dramatically impact profitability.

Dog #3 says: "Restaurant economics are brutal. The four-wall EBITDA is your only real signal — strip out corporate overhead and see if the actual restaurant makes money."

Restaurant models are unit economics at heart. Build one store P&L perfectly, then layer in new unit opening timelines. Prime costs (food + labor) as a combined percentage is the survival metric — it needs to stay below 65% to leave room for occupancy and overhead.

Dog #4 says: "SaaS models are ARR waterfall models. New ARR − Churned ARR = Net New ARR. Simple. Until your churn cohorts mature and you realize your retention is terrible."

SaaS models are driven by the ARR waterfall: opening ARR + new bookings + expansion − contraction − churn = ending ARR. The unit economics (LTV / CAC) determine whether it's worth spending on growth. Rule of 40 is the summary metric that tells investors whether growth + profitability balance.

Dog #5 says: "Capacity utilization is everything. A plant running at 60% is burning cash. The same plant at 85% prints money. That's the non-linearity of manufacturing."

Manufacturing models start with volume × price, then build COGS from the ground up: raw materials + direct labor + overhead absorption. Fixed overhead gets allocated to units — so volume swings create enormous margin volatility even if prices hold steady. A 10% volume drop can cut gross margin in half.

Dog #6 says: "Banks are not normal companies. Revenue is net interest income — a spread. The model inverts: assets generate revenue, liabilities fund them. Get the balance sheet right first."

Banking models start with the balance sheet, not the income statement. Loans and investments earn interest (assets); deposits and debt cost interest (liabilities) — the spread is NIM. Add fee income, subtract credit losses and operating expenses, apply tax. Every assumption has a regulatory constraint attached.

Dog #1 says: "Healthcare models are payor mix models in disguise. Medicare pays 80 cents on the dollar, Medicaid pays 60, commercial pays 110. Who's in your waiting room determines your margin."

Hospital models build revenue from patient volume × payor mix × reimbursement rates. Gross charges are fictional — net revenue depends entirely on the payer contract and bad debt / charity care write-offs. Always model net patient revenue, never gross.

Dog #2 says: "Regulated utilities are bond-like — predictable, boring, capex-heavy. The model is: rate base × allowed ROE = net income. Simple once you accept the regulator is your silent partner."

Utility models split between regulated (rate-of-return) and merchant (market-price). Regulated utilities earn a set return on their capital base — model capital spending and rate cases. Merchant generation models commodity prices, spark spreads, and capacity market revenues. They look like completely different businesses because they are.

Dog #3 says: "Pharma models are probability-weighted NPV models. Your pipeline is worth the sum of (peak sales × probability of approval × NPV factor) for each drug. The rest is details."

Pharma models combine two very different businesses: the cash-cow marketed portfolio and the binary-outcome R&D pipeline. Modelers must probability-weight pipeline assets and sensitize against patent cliffs and generic entry timing. Gross-to-net adjustments (rebates, chargebacks) can reduce sticker price by 30–50%.

Dog #4 says: "Telecom models are subscriber cohort models with big capex hangovers. The question is always: is ARPU growing faster than churn? If yes, build the towers. If no, you're just a declining annuity."

Telecom models track subscriber counts × ARPU minus churn-driven losses. The capital intensity of network buildout (spectrum, towers, fiber) makes FCF timing critical — EBITDA looks great, but capex devours it for years before the network fully monetizes.

Dog #5 says: "Marketplace models are GMV × take rate. The hard part is that your take rate compresses as you scale — buyers and sellers get leverage. Model the take rate as a declining function of GMV."

Marketplace models track gross merchandise value (GMV) — the total value of transactions — then apply the platform's take rate to get revenue. The classic tension is between growing GMV (requires low prices / subsidies) and improving take rate (requires more value-add services).

Dog #6 says: "Insurance is a float business. You collect premiums today, pay claims later. The underwriting result tells you if you're profitable on the insurance side. The investment income tells you if you're profitable on the float."

Insurance models separate underwriting (premiums vs. claims + expenses) from investment income (float). Combined ratio is the core metric — above 100% means you're losing money on underwriting and need investment income to compensate. Below 100% means you get paid to hold other people's money.

Dog #1 says: "Media models are audience × monetization rate. Whether it's CPM, CPC, or subscription, you need to model the audience first — and audiences are the hardest thing to forecast."

Media models bifurcate into advertising revenue (impressions × CPM or clicks × CPC) and subscription / licensing. The secular shift from ad-based to subscription revenue is the dominant trend — model both tracks and the mix shift over time. Content cost is the swing variable that can destroy an otherwise-healthy margin profile.

Dog #2 says: "Professional services is a people model. Headcount × bill rate × utilization = revenue. If utilization drops below 70%, you're burning cash. If it's above 85%, your people are burning out."

Services firms model billable headcount × utilization rate × average bill rate. The pyramid structure (partners leverage juniors) determines margin — model partner-to-staff ratios and their respective comp as the key COGS driver. Attrition is an operating risk, not just an HR metric.

Dog #3 says: "EdTech models look like SaaS but aren't. Content has high upfront cost and low marginal cost. CAC is high (education is a grudge purchase). Completion rates tell you if you'll get word-of-mouth."

EdTech models blend subscription, course-by-course, and B2B enterprise sales. Content is a depreciating asset — a course launched in 2020 needs refresh spend by 2025. Model content library value alongside subscriber growth and treat content amortization as a real cost, not a non-cash afterthought.

Dog #4 says: "Gaming models are live-service economics: you launch a game and it's a platform. DAU × ARPDAU is the engine. The question is how many of your DAUs are paying players — and how much."

Mobile gaming models use ARPDAU × DAU for live-service titles. The economics pivot on conversion from free to paying — top spenders ("whales") generate the majority of IAP revenue, so whale retention is existential. D1 / D7 / D30 retention curves compound fast: a 1% improvement in D30 retention can double LTV.

Dog #5 says: "Real estate is a cap rate model. NOI / Cap Rate = Value. If cap rates expand by 50bps and your NOI doesn't grow, you just lost 5–8% of asset value. That's the rate risk nobody talks about at the pitch."

Real estate models layer three analyses: the property-level NOI model, the project-level development return, and the portfolio / REIT model. The discount rate and cap rate assumptions drive valuation far more than operating assumptions — sensitivity tables on cap rates are mandatory, not optional.

Dog #6 says: "A startup model has one job: show how long until you die, and what it takes to not die. Runway × burn rate × milestones. Everything else is fiction dressed up as planning."

Startup models are burn-rate models with aspirational upside. Build the bottoms-up cost structure honestly — headcount drives 60–80% of costs. Then layer in revenue scenarios. The key deliverable is a runway analysis showing how different revenue outcomes change your cash-out date and next fundraise timing.