Build a Clean Assumptions & Driver Sheet That Runs the Financial Model
A good financial model rarely breaks because the income statement formula is difficult. It breaks because one small assumption - price per unit, collection days, churn, utilization, tax rate - is buried in Sheet 7, hardcoded twice, and quietly drives three different answers.
- The assumptions and driver sheet is the model control panel: all key inputs live in one place and flow into the forecast.
- Separate inputs from calculations: blue cells for manual inputs, black cells for formulas is a common banking convention.
- Good drivers are operational, not decorative: revenue should connect to volume, price, customers, utilization or take rate wherever possible.
- Every material assumption needs a source: historical average, management guidance, industry benchmark, contract term or analyst judgement.
- Scenarios should change driver cells, not output cells: Base, Upside and Downside must flow through the model mechanically.
- Use sanity checks: growth rates, margins, working-capital days, leverage ratios and balance-sheet checks should flag unrealistic outputs.
- The common trap is hardcoding forecast numbers directly into financial statements instead of building them from drivers.
Big Picture - The Driver Sheet Is the Model's Control Panel
Think of the assumptions sheet as the cockpit of a financial model. The financial statements are the aircraft's movement; the driver sheet is where you set speed, altitude, fuel and route. If the cockpit is clean, the model becomes easy to audit, update, scenario-test and explain.
Core Explanation - What Goes Into a Strong Assumptions Sheet
The big idea is simple: do not forecast financial statements first; forecast the business first. A revenue line is not just "sales grow 12 percent". It is customers, volume, price, mix, retention, occupancy, utilization, transaction value, take rate or capacity - depending on the business.
A strong assumptions sheet usually has six zones:
Driver-Based Forecasting - The Clean Mental Model
A driver-based model links financial results to the economics of the business. The test is: if the CEO says "we are adding 200 stores" or "retention improved", your model should know where that changes revenue, cost, working capital and capex.
Types of Assumptions You Must Separate
Interviewers like this topic because it tests whether you can model with discipline. Use this classification to avoid mixing a strategic assumption with an accounting policy or a financing input.
How to Build the Sheet - A Practical 7-Step Process
Build the sheet before you build the forecast schedules. Otherwise you will end up retrofitting logic into a messy workbook.
Mini Worked Example - Turning Assumptions Into Revenue
Here is a tiny illustrative model for a beverage distributor. The point is not the industry number; the point is the modelling logic.
If you had simply typed "revenue grows 21.3 percent", the number would be harder to defend. With drivers, you can explain the growth as outlet expansion, higher throughput per outlet and better realization.
Assumption Sheet Quality Metrics - What to Track
A driver sheet is not "good" because it looks neat. It is good when it is traceable, flexible and hard to break. Use these checks while reviewing your own workbook.
Definitions - Say These Cleanly
Assumption: A forecast input based on historical evidence, guidance, benchmark data or analyst judgement.
Driver: An operational variable that causes a financial line item to change.
Scenario: A coherent set of assumptions representing a possible future case.
Sensitivity analysis: A technique that measures how an output changes when one input changes, holding other inputs constant.
Scenario Design - How Base, Upside and Downside Should Work
Scenarios should be coherent stories, not random percentages. A Downside case for a retailer may combine lower footfall, weaker gross margin and slower inventory turns. An Upside case may combine higher volume, better mix and operating leverage. The mistake is changing EBITDA directly - that hides the business reason.
Case Study - Varun Beverages: Modelling Growth Through Drivers
Varun Beverages is a useful Indian example because its model is naturally driver-based: territories, volumes, realization, seasonality, distribution reach, margins and capex all interact.
Situation: Varun Beverages, one of PepsiCo's major franchise bottlers, operates in a business where reported revenue is the visible output, but the real engine is physical volume, distribution reach, product mix, pricing and capacity. For an analyst, simply assuming a revenue growth percentage would miss how the business actually works.
The modelling move: A stronger model would build revenue from cases sold and realization per case, then link margins to input costs, operating leverage and mix. Capex would not be an afterthought; it would connect to capacity expansion, coolers, plants and route expansion. Working capital would reflect inventory build-up, receivables and payables around a seasonal beverages business.
Outcome or lesson: The primary driver is volume-led expansion through territories and distribution depth. Supporting drivers include pricing, portfolio mix, manufacturing scale, cold-chain assets, execution discipline and capex timing. The strategic lesson: for operating businesses, a driver sheet turns a vague growth story into a testable financial model.
How AI Changes Building the Assumptions & Driver Sheet
AI does not replace modelling judgement. It makes the research, audit and challenge process faster - if you still own the final assumptions.
- Faster source extraction: Tools can scan annual reports, investor presentations and earnings-call transcripts to pull management guidance, segment commentary, capex plans and risk factors into an assumption log.
- Driver discovery: An LLM can suggest potential drivers by business model - for example, AUM, yield, credit cost and cost of funds for an NBFC; occupancy, ARR and RevPAR for hotels; stores, footfall and average bill value for retail.
- Model audit support: AI-assisted spreadsheet tools can help identify inconsistent formulas, hardcoded numbers, broken links and unusual changes across forecast periods, but they must be reviewed by the modeller.
Load the company annual report, latest investor presentation and your assumption sheet into NotebookLM. Ask: "List all management statements that support or challenge my revenue, margin, capex and working-capital assumptions, with source references." Then use ChatGPT or Claude to generate downside-case questions an interviewer may ask.
Interview Relevance
"If you were building a financial model from scratch, how would you design the assumptions and driver sheet?"
Use one industry-specific example in your answer. For a quick-commerce company, say revenue could be driven by orders, average order value, take rate and delivery fee, while costs link to rider cost, dark-store rent, fulfilment cost and marketing spend.
Common Mistake
The biggest mistake is typing forecast outputs directly into the income statement and calling them assumptions. It costs candidates because it shows poor modelling discipline and makes scenarios impossible to audit. One-line fix: put every key input on the assumptions sheet and let formulas drive the statements.
What to Revise Next
Once you can build the assumptions control panel, move to the engine it drives: The Three-Statement Model, Built From a Blank Workbook. After that, revise Making the Model Balance: Debt Schedules, Cash Sweep & Circularity, where assumptions interact with financing, cash flow and balance-sheet mechanics.