Financial Model Research Methodology
Skip to content
1Understand the ideaCustomer, offer, transaction, delivery, capacity, and cash cycle.
2Design the modelBusiness-model logic, statements, schedules, drivers, and KPIs.
3Research the inputsDefinitions, sources, units, geography, period, and ranges.
4Test the systemReconciliations, sensitivities, downside cases, and limitations.
Research methodology
From a business idea to a financial model you can inspect.
Every business earns, spends, invests, and converts activity into cash differently. We research the business-model architecture first, then select the reports, KPIs, revenue and cost drivers, and numerical assumptions the model needs.
The core rule: we do not begin with a generic spreadsheet or a precise-looking number. We begin with how the business operates, then make the model logic and evidence traceable.
Business model
Choose the right economic logic
SaaS, marketplace, restaurant, retail, service, and location models require different mechanics.
Architecture
Choose reports, schedules, and KPIs
The model structure follows the operating cycle and the decisions it must support.
Inputs
Research drivers—not decorative averages
Volume, price, staffing, cost, timing, capital, and financing assumptions need defined evidence.
Validation
Trace assumptions into cash and results
Linked statements, scenarios, checks, and sensitivities expose how the model behaves.
Scope
One method for researching the model structure and the numbers inside it.
For each new or materially updated financial model, Financial Models Lab researches the underlying business idea before finalizing the workbook architecture. Older products and legacy editorial pages may have been produced under earlier workflows and may not yet contain the same research records, definitions, or review markers.
The work has two linked parts. Model-design research determines the appropriate business-model logic, operating drivers, statements, schedules, KPIs, scenarios, and timing relationships. Assumption research evaluates the numerical inputs used to illustrate or test that structure, including volume, price, staffing, cost, working capital, capital expenditure, financing, and timing.
The method also covers editorial claims about startup costs, operating costs, prices, wages, market context, profitability, break-even, owner compensation, and similar planning metrics. Neither the structure nor the researched inputs turn a scenario into a personalized forecast or guaranteed result.
Commercial link
Financial Models Lab sells templates and services. Research content may link to those products. Own-product links do not count as independent evidence for an industry fact or market estimate.
Definitions first
We define the number before we collect it.
Many apparent source conflicts are actually definition conflicts. A business can have high revenue and low owner income; a profitable month can still have negative cash flow.
- Business-model archetype
- The economic pattern that best describes how the business creates value, charges customers, delivers the offer, incurs costs, and receives cash. One business can combine more than one archetype.
- Model architecture
- The connected set of inputs, calculations, operating schedules, financial statements, KPIs, scenarios, and checks required for the stated planning purpose.
- Revenue driver
- An operational variable that explains revenue formation, such as active subscribers, transactions, average check, units sold, occupied capacity, billable hours, price, or take rate.
- Key performance indicator
- A defined measure used to monitor an operating or financial relationship. A useful KPI must match the business model and state its formula, period, and unit.
- Gross sales or bookings
- The value of transactions recorded before specified returns, allowances, discounts, cancellations, or other adjustments. It is not automatically recognized revenue and must state whether taxes or other pass-through amounts are included.
- Revenue
- Income recognized from ordinary business activities under the stated reporting basis, generally net of returns, allowances, discounts, and amounts collected for third parties. If a source reports gross sales or bookings instead, we label that measure explicitly.
- Gross profit
- Revenue less the direct cost of producing or delivering the good or service. Cost classification can differ by industry.
- Operating profit
- Profit after operating expenses but before items such as financing and tax, subject to the stated accounting definition.
- Owner income
- An umbrella term that may include salary, wages, draws, distributions, or other economic benefit depending on legal and tax structure. We state which components are included and do not equate owner compensation with distributable cash.
- Cash flow
- Cash movement over time, including timing effects that may not appear in accounting profit in the same period.
- Startup funding need
- The cash required for a stated scope and runway, which may include assets, deposits, pre-opening expenses, opening inventory, working capital, and contingency. Owner living costs and financing offsets are shown separately unless expressly included.
- Break-even
- The point at which a defined contribution or profit measure covers a defined cost base. It is not necessarily positive cash flow.
- Market size
- A measured or estimated pool of economic activity within a defined product category, geography, and period—not automatic demand for one business.
- Forecast
- A scenario derived from assumptions. It is not a known future result, promise, or valuation opinion.
Business-model fit
The same spreadsheet logic does not fit every business idea.
Before choosing formulas, we map who pays, what is sold, the unit of sale, purchase frequency, delivery method, operating capacity, cost behavior, and cash timing. “Best fit” means the most defensible representation of that business idea—not that one business model is universally superior.
| Business-model archetype | Core revenue logic | Illustrative drivers and KPIs | Model implications |
|---|---|---|---|
| Recurring subscription / SaaS | Active customers × recurring price, adjusted for acquisition, expansion, contraction, and churn. | MRR or ARR, ARPU, new customers, churn, retention, CAC, and gross margin. | Subscriber or cohort schedules, acquisition and support costs, renewal logic, and monthly-versus-annual billing cash timing where relevant. |
| Marketplace / platform | Transaction value or volume × take rate, plus applicable listing, subscription, or service fees. | Buyers, sellers, transactions, GMV, average order value, take rate, repeat rate, refunds, and contribution margin. | Two-sided activity, commissions, seller payouts, incentives, refunds, payment fees, and settlement timing. |
| Restaurant / hospitality | Covers, orders, or occupied units × average check or rate across dayparts and channels. | Seats, turns, covers, occupancy, average check, food cost, labor percentage, delivery mix, and revenue per available unit. | Capacity and daypart logic, menu or channel mix, inventory and waste, staffing, rent, equipment, and opening investment. |
| Retail / ecommerce | Traffic or footfall × conversion × units per order × price, net of returns and discounts. | Conversion, average order value, gross margin, return rate, inventory turns, CAC, and channel performance. | Inventory purchasing, fulfillment, returns, marketplace or payment fees, marketing, and working-capital timing. |
| Service / capacity / location | Leads, jobs, billable hours, members, or occupied capacity × price. | Lead conversion, utilization, billable hours, revenue per employee or location, labor efficiency, and capacity. | Staffing and scheduling, capacity constraints, travel or material costs, utilization ramp-up, and multi-location rollout where relevant. |
| Manufacturing / asset-heavy | Units produced and sold × price, constrained by capacity, yield, downtime, and production lead time. | Throughput, utilization, yield, scrap, unit cost, contribution margin, inventory days, and return on invested capital. | Raw material, work-in-progress, and finished-goods inventory; production labor and overhead; maintenance, capital expenditure, depreciation, and debt. |
Hybrid businesses: a company may combine subscriptions, transactions, products, services, advertising, or locations. We model each material revenue stream separately and connect them only where the operating relationship is explicit.
Model blueprint
How a business idea becomes a workbook architecture.
The blueprint is the bridge between business research and spreadsheet construction. It defines what the model must explain before numerical assumptions are collected.
Define the decision and scope
State the intended user and decision, forecast horizon, geography, currency, reporting frequency, and the boundaries of the business being modelled.
Select or combine archetypes
Choose the economic pattern that best represents each material revenue stream and document why it fits the business idea.
Map operations, capacity, and timing
Identify the operating units, customer journey, fulfillment steps, staffing constraints, seasonality, ramp-up, and cash-conversion cycle.
Design the revenue engine
Express revenue through observable drivers such as customers, transactions, volume, capacity, price, frequency, mix, churn, or take rate instead of an unexplained growth percentage.
Design the cost, capital, and cash engines
Separate variable and fixed costs; map labor, inventory, working capital, capital expenditure, financing, and tax logic where they are relevant to the stated purpose.
Select statements and supporting schedules
Choose the profit and loss statement, cash flow, balance sheet, and supporting schedules—such as staffing, inventory, debt, and fixed assets—needed to connect operations to financial results.
Select KPIs and scenario levers
Use business-specific metrics with defined formulas, periods, and units, then identify the assumptions that should change across low, base, high, or named scenarios.
Design the validation plan
Specify reconciliations, sign and balance checks, roll-forwards, scenario integrity tests, sensitivities, and usability checks before final review.
Product scope controls: not every workbook needs every statement, schedule, or KPI. The individual product page and delivered file define what is included; this methodology explains how relevant components are selected for new or materially updated models.
Evidence hierarchy
Different sources answer different questions.
We do not treat an operator comment, vendor quote, trade survey, and official dataset as interchangeable. Weight depends on the claim, methodology, coverage, age, and independence of the source.
Tier 1 · Primary public evidence
Official statistics, regulators, laws, filings, and original datasets
Preferred for wages, inflation, business counts, economic activity, regulatory requirements, public-company benchmarks, and definitions.
Tier 2 · Independent research
Academic work, established research organizations, and method-documented trade reports
Useful for industry context, operating benchmarks, survey findings, and topics not covered at sufficient detail by public data.
Tier 3 · Current market evidence
Vendor pricing, job listings, property listings, menus, rate cards, and quotations
Useful for observable current costs and prices. Commercial incentives, taxes, geography, package differences, and temporary promotions must be considered.
Tier 4 · Operator evidence
Interviews, case studies, practitioner communities, and firsthand accounts
Useful for workflow, failure modes, edge cases, and reality checks. Anecdotes are not used alone to establish a universal earnings figure.
Open the Data Sources page for named examples, intended use, and known limitations.
Numerical assumption research
How model inputs are researched for the business idea.
Once the architecture is defined, each material input needs a meaning, unit, period, geographic basis, evidence trail, and role in the model.
Build the assumption register
List each material input from the blueprint, its driver, definition, unit, geography, period, tax and gross-versus-net treatment, and the output it affects.
Build a source plan
Identify which primary data could answer the question, where industry evidence is needed, and which current prices or operator accounts can test practical fit.
Collect and retain source notes
Record the publisher, page or dataset, URL, publication/reference period, access date, relevant definition, geographic coverage, and material limitation.
Normalize before comparing
Align currencies, time periods, units, inflation basis where appropriate, taxes, channel, business scale, employment status, and gross-versus-net definitions.
Translate evidence into model drivers
Convert source evidence into the exact inputs and formulas required by the model, document the bridge, and check that related costs or revenues are not counted twice.
Triangulate and set ranges
Compare relevant sources, explain disagreements and outliers, and establish low, base, and high values or named cases from a stated basis.
Test and review the linked model
Verify arithmetic, statements, sensitivities, cash timing, seasonality, missing costs, downside behavior, scenario integrity, and the limitations attached to the inputs.
Minimum evidence rule: a material quantitative claim should have at least one directly relevant primary source or documented current observation. If no suitable primary source exists, use two independent relevant sources where reasonably available and explain the exception. Do not turn one anecdote into a typical result. When credible sources conflict, reconcile definitions and scope; otherwise show the disagreement, lower confidence, widen the range, or omit the claim rather than average incompatible figures.
Demand and volume
How much activity can occur?
Customer counts, traffic, conversion, capacity, occupancy, transactions, units, seasonality, and ramp-up.
Pricing and revenue
How does activity become revenue?
Price lists, menus, subscriptions, contracts, take rates, product mix, discounts, cancellations, and returns.
People and productivity
What work and staffing are required?
Wages, roles, hiring timing, utilization, output per employee, scheduling, and capacity per team or location.
Operating costs
What changes with scale—and what does not?
Materials, platform fees, rent, utilities, logistics, marketing, insurance, software, maintenance, and waste.
Capital and working capital
When does cash enter or leave?
Equipment, deposits, inventory days, receivable and payable timing, operating reserve, replacement cycles, and opening investment.
Financing, tax, and regulation
Which external rules shape cash?
Debt terms, financing fees, taxes, licenses, and required compliance costs, with qualified review where jurisdiction-specific advice is needed.
A researched input may still be unsuitable for a specific company, location, or date. Default and sample values are editable starting assumptions—not predictions, guarantees, or universal “typical” numbers.
Ranges and calculations
Evidence becomes an editable range—not a hidden “correct” number.
A range may come directly from a credible dataset, from multiple current observations, or from a transparent component calculation. We retain the low, base, and high basis, source date, definition, formula, and linked model driver rather than hiding uncertainty behind one average.
| Label | What it means | Minimum context | What it does not mean |
|---|---|---|---|
| Source-reported value | A value reproduced from a cited source while preserving its stated status, such as observed, estimated, modelled, revised, or provisional. | Publisher, original status, period, geography, unit, and link. | Independent verification or applicability to every business. |
| Derived estimate | A calculation using cited evidence and stated assumptions. | Formula, components, scope, and sensitivity. | That the source itself published the result. |
| Planning range | A low-to-high band intended for scenario planning. | Drivers, exclusions, geography, and basis for bounds. | A guarantee that actual results will fall inside it. |
| Example scenario | An illustrative set of assumptions used to explain mechanics. | Clearly labeled inputs and calculation period. | A benchmark, typical result, or prediction. |
| Operator account | A firsthand report used for context or a reality check. | Role, business context, date, and whether identity is public. | Statistically representative evidence. |
Example of a transparent derived range
Startup cash required = one-time opening costs + pre-opening expenses + opening inventory + operating reserve − committed financing available at launch.
Each component should have its own source or assumption. The range should state whether owner living costs, taxes, deposits, contingencies, debt service, and working capital are included. Changing the reserve period should visibly change the result.
Changing a material price, volume, cost, staffing, capacity, or working-capital assumption should flow visibly through the relevant schedules, financial statements, cash position, KPIs, and scenarios.
Confidence and limits
Uncertainty is part of the result.
When useful, a page may summarize evidence strength. Confidence refers to the support for the published range—not confidence that a reader will achieve it.
Higher confidence
Multiple current, independent sources align
Definitions match, geographic coverage is relevant, sample or administrative data is substantial, and calculations are directly reproducible.
Moderate confidence
Evidence is credible but incomplete
Sources broadly agree, but detail, recency, geography, sample size, or definition creates meaningful uncertainty.
Lower confidence
Evidence is sparse or highly variable
Current data is limited, sources conflict, outcomes are operator-specific, or the estimate relies heavily on observable market examples.
No confidence label removes business risk. Local demand, competition, execution, financing, regulation, timing, and luck can move actual outcomes materially.
Review and corrections
A review date is not a guarantee that nothing changed yesterday.
Materially updated research should display a review date. Pages and models are candidates for review when the business-model architecture changes, a KPI definition changes, a source releases new data, a law or program changes, a material link breaks, a credible correction is submitted, or an editorial review identifies stale assumptions.
If you find a likely error, use the contact page. Include the URL, exact claim, why it appears wrong, and a credible supporting source. We distinguish silent typo or link fixes from material changes to a number, conclusion, or methodology.
For authorship, AI-assisted workflows, own-product references, sponsorships, and correction records, see our Editorial Standards.
Responsible publisher: Financial Models Lab · Methodology draft version 1.0 · Prepared August 2, 2026 · Effective only after internal approval.
Inspect the evidence
See the source types behind model design and assumptions.
Open the filterable register to review intended uses, geographic coverage, primary-source status, and known limitations.