SaaS operating model
Build a subscription business from sales capacity to ARR to a quarterly P&L, reconcile cash to revenue through deferred revenue, and read the unit economics an investor will ask for.
Technology companies are modelled from the objects their business runs on rather than from a growth rate. A subscription software company sells contracts: its revenue is annual recurring revenue moved by new sales, expansion and churn, constrained by the capacity of its sales team, billed ahead of service, and judged on retention and what a customer costs to win. A marketplace sells nothing of its own: buyers spend with sellers, the platform keeps a share of that gross merchandise value, and it is judged on whether buyer cohorts keep spending, what each order leaves after its costs, and when a buyer repays its acquisition cost.
By Surojit Chakraverti, founder of L3VLUP and an investor running a long-short healthcare and technology equities strategy.
On this subject: technology.Updated 2 October 2026
Build a subscription business from sales capacity to ARR to a quarterly P&L, reconcile cash to revenue through deferred revenue, and read the unit economics an investor will ask for.
Build GMV from buyer cohorts with retention and frequency by tenure, take it to net revenue through the take rate and incentives, to contribution per order and EBITDA, and judge a buyer’s lifetime value against its acquisition cost.
Build a payments company from processed volume by channel and region, take it through gross receipts and the pass-through costs to net revenue and the net take rate, to costs, EBITDA and free cash flow, roll the settlement float forward, and value the company on forward multiples and a DCF.
Build an integrated telecom operator from its customers and its networks: postpaid and prepaid subscribers rolled forward by gross additions and churn, ARPU by segment, handset revenue kept apart from service revenue, a fibre network from homes passed to connections as penetration ramps, costs to EBITDA and then to EBITDAaL after leases, capital spending split between maintenance and the fibre build, a spectrum auction amortised over its licence, then leverage, equity free cash flow, dividend cover and a DCF checked against a peer multiple.
Build a foundry from its fabs: capacity by node added a year after the spending, utilisation through a cycle, revenue as wafers times price per wafer, yield as a cost, fixed and variable cost of revenue, a vintage-by-vintage depreciation waterfall, and free cash flow valued in a DCF, then read what a downturn year does to gross margin.
Rebuild a company’s own non-GAAP measures from its GAAP lines, model a subscription business as subscribers times what each pays, carry it five years in three labelled cases, and read what the forecast is worth at a chosen discount rate without mistaking it for a view on the shares.
Build a marketplace from nights and the price of a night, derive a take rate the company never files for a year, tell bookings, revenue and cash apart in every row, keep customer funds out of company cash and value, and read what the forecast is worth without mistaking it for a view on the shares.
Model a two-segment streaming business from its own KPIs, keep a segment reclassification from manufacturing growth, separate share-price-driven items from earning power, carry the forecast five years in three labelled cases, and read what it is worth without mistaking it for a view on the shares.
Model a payments platform from volume and take rate, keep volume, gross and net revenue from being confused, read a take rate as the product of a mix and a set of pillar rates, measure operating leverage against a hiring plan, value the business without counting merchants’ money or the interest on it as the company’s, and read the value without mistaking it for a view on the shares.
Other areas: Core modelling · M&A and private equity · Credit and restructuring · Real estate · Infrastructure and project finance · Healthcare · Financial institutions · All 57 models
Start with the marketplace economics lab, which works one order and one buyer cohort by hand. Then open the marketplace model to see cohorts become GMV, net revenue and EBITDA over five years. The SaaS model is the subscription counterpart: switch it to its downturn case and watch net retention, the magic number and the Rule of 40 move together. The payments model is the third shape: processed volume, gross receipts and net revenue kept apart, with the net take rate in basis points as the number that matters. The semiconductor foundry model is the capital-heavy shape: wafer capacity bought a year ahead, utilisation through the cycle, and a cost base so fixed that its downturn case takes ten points off gross margin. The telecom operator model is the subscription business with a network under it: subscribers rolled forward by gross additions and churn, a fibre build whose spending comes years before its revenue, and a dividend tested against free cash flow after leases and spectrum. The unit economics and SQL cohort labs drill the customer-level arithmetic both models share. Each model has a starter workbook that leaves one schedule for you to build.
Each schedule is one reusable calculation, explained on its own page with the rows it occupies in every model that uses it.
Technology schedules
Sales capacity
Reps, ramp, quota: the new ARR a team can close
Technology schedules
ARR roll-forward
Opening, new, expansion, churn, closing
Technology schedules
Bookings, billings and deferred revenue
Cash before revenue, and the balance in between
Technology schedules
SaaS profit and loss
Gross margin, sales and marketing, R&D, G&A, cash
Technology schedules
Unit economics
CAC, payback, LTV, retention, magic number, burn multiple
Technology schedules
Cohort GMV
Buyer cohorts, retention, frequency, orders, GMV
Technology schedules
Take rate
From GMV to net revenue
Technology schedules
Contribution margin
What an order leaves after the costs it brings
Technology schedules
Buyer economics
One cohort: GMV retention, LTV to CAC, payback
Technology schedules
Processed volume: TPV by channel and region
TPV, transactions and average transaction value
Technology schedules
Take rate and pass-through: gross receipts to net revenue
Interchange, scheme fees, acquiring costs, net take rate
The statements
Operating costs and margins
Cost of sales, opex, EBITDA
Technology schedules
Settlement balances and the float
Due from schemes, due to merchants, the float
Operating schedules
Capex and depreciation (PP&E roll-forward)
Opening, plus capex, less depreciation, closing
The statements
Tax
Tax on profit, and tax on operating profit
Valuation and returns
Enterprise value to equity value
Debt off, cash on, per share
Valuation and returns
Discounting and terminal value
Discount factors, two terminal methods, reconciled
Valuation and returns
Cost of capital (WACC)
CAPM, after-tax debt, target weights
Valuation and returns
Sensitivity tables
Two assumptions at once, without a data table
Operating schedules
Mobile subscriber roll-forward: gross adds, churn, net adds
Postpaid and prepaid bases, gross adds, churn, net adds
The statements
Revenue build
Where the top line comes from
Operating schedules
Fibre build: homes passed, penetration and connections
Homes passed, a penetration ramp, capex per home
Operating schedules
Spectrum licences: auction payments and amortisation
An auction paid in one year, amortised over the licence
Operating schedules
Working capital schedule
Receivables, inventory and payables in days
Financing schedules
Debt schedule
Tranches, interest, amortisation and the sweep
Technology schedules
Wafer capacity, utilisation and shipments
Fab capacity by node, the spending lag and the cycle
Technology schedules
Fab cost structure: yield, fixed costs and margin
Variable and fixed cost of revenue, and operating leverage
The statements
Cash flow statement
Indirect method, closing on balance-sheet cash
Unit Economics · 12 min
Build an LTV:CAC you could defend under questioning, and answer payback as its own question.
Product Metrics Tree · 12 min
Name the right north star for a product shape, and the guardrail your change would break.
Marketplace Economics · 12 min
Turn GMV into net revenue and contribution per order, read a cohort’s GMV retention from retention and frequency, and find the payback year on a buyer.
SQL Cohort Drill · 25 min
Write the cohort, retention and rolling-active queries interviews ask for, and read each result without misreading it.
Name That Multiple · 3 min
Place a business in the right EV/EBITDA band from its quality alone, and say what drives it.
DCF Builder · 15 min
Build a DCF by hand from revenue to a share price, know how much of it sits in the terminal value, and name the one assumption the answer hinges on.
Covenant Headroom · 10 min
Compute the three maintenance ratios from a certificate, find the binding test and its EBITDA cushion, and say whether a given downside breaches.
Depreciation Waterfall · 10 min
Forecast depreciation from a capex plan rather than a ratio, read a capex-to-depreciation multiple, and say which part of the spend is buying growth and which is only standing still.
EV to Equity Bridge · 12 min
Walk from enterprise value to a price per share without dropping a claim: net debt, pensions, minorities, leases and options.
Reverse DCF · 10 min
Read off the growth a share price already assumes, and say whether that is plausible before anyone asks.
Three-Statement Linker · 8 min
Walk any event through all three statements in the right order, with the right signs, and prove the balance sheet balances rather than asserting it.
WACC Builder · 12 min
Build a discount rate you can defend line by line: beta, equity risk premium and cost of debt, each from a source you can name.
Beta Refresher · 10 min
Estimate a beta, see how far one regression lands from the truth, and know when to use raw, adjusted, unlevered or bottom-up.
IRR by Eye · 3 min
Say the IRR for any multiple and hold before the interviewer finishes the sentence, and the multiple for any IRR.
LBO Modelling Test · 60 min
Sit the one-hour test under a clock, in the browser or in your own workbook, and get every line of the build marked.
Capital Stack Challenge · 15 min
Finance a real buyout from revolver to equity, hold it through a shock, and explain every layer against the deal the sponsor actually did.
Optimal Capital Structure · 15 min
Explain why more debt stops helping: watch coverage, rating and spread move as leverage rises until WACC turns.
Minority stakes in companies that already work and need capital to scale. Between PE and VC.
Early-stage minority investing. Judgement about people and markets, far less modelling.
How to get in and prepareCover a sector, publish a view, defend it to clients. The best writing training on the map.
How to get in and prepareBuy companies for an operator. The modelling is banking; the judgement is not.
How to get in and prepareIn-house consulting with implementation attached. Half analysis, half getting it done.
How to get in and prepareWhat the companies these models describe tell their own investors about growth, margins and long-term bets.
Shareholder letter · 1998
Amazon's 1997 letter to shareholders
Amazon's first shareholder letter sets out a long-term, cash-flow-first philosophy that still frames how investors judge high-growth companies.
Shareholder letter · 15 Oct 2025
Microsoft annual report letter to shareholders, fiscal 2025
How the largest software company presents the AI platform shift to investors: datacentre build-out, Copilot adoption and the culture meant to deliver it.
IPO founder letter · 29 Apr 2004
Google's 2004 IPO founders' letter, "An Owner's Manual"
The founding text on dual-class shares, refusing earnings guidance and funding long-shot bets; essential background for founder-control debates in IPOs.
A SaaS model starts from contracts: annual recurring revenue moved by new bookings, expansion and churn, with revenue recognised over the contract. A marketplace model starts from buyers and orders: cohorts of buyers, how often they order, the value of each order, and the share of it the marketplace keeps. One is judged on net revenue retention and CAC payback; the other on GMV retention, take rate and contribution per order.
For software: ARR, net and gross revenue retention, CAC payback, LTV to CAC, the magic number and the Rule of 40. For marketplaces: GMV against revenue, the net take rate, contribution per order, cohort GMV retention and buyer payback. The labs drill both sets, and each model page lists the mistakes a reviewer looks for.
Because an average hides whether the business is getting better or worse. A blended retention rate can hold steady while every new cohort retains worse than the last, and a cohort view shows it. Both models build revenue from customers or buyers acquired in each period for that reason.