Here is the answer up front. When an investor, or a lender, or an acquirer, opens your financial model, they are checking seven things in roughly this order: can they understand the structure in five minutes, are the assumptions labeled and defensible, does revenue build from real drivers rather than a growth percentage, do costs scale believably, do the unit economics work, does the cash runway math hold under a downside case, and does the forecast connect honestly to your historical numbers. A model that passes all seven rarely wins the deal by itself. A model that fails any two of them reliably loses it, because the model is being read as a proxy for how you think.
Celeste Business Advisors builds and repairs financial models inside our FP&A and fractional CFO work, and we have sat on both sides: preparing models for capital raises and sales, and reviewing them for buyers. This guide is the reviewer's checklist, written from the reviewing side of the table.
1. Structure They Can Navigate in Five Minutes
A reviewable model separates inputs, calculations, and outputs: one assumptions tab where every number that can be changed lives, calculation tabs with formulas and nothing hardcoded, and output tabs presenting the three statements and a summary. Reviewers test this within minutes by tracing one number, this quarter's revenue, back to its sources. If the trace dead-ends in a typed constant buried in a formula, trust drops immediately and everything after is read skeptically. The three-layer structure is the same one we describe in turning spreadsheets into strategy dashboards; it exists for your own decision-making first, and it happens to be exactly what diligence rewards.
2. Assumptions That Are Labeled, Sourced, and Defensible
Every assumption should be visible, named, and attached to a reason: conversion rates from your actual funnel history, pricing from your actual contracts, market growth from a named source. "Conservative" is not a defense; evidence is. The tell reviewers look for is asymmetry, models where every assumption happens to break optimistic. One honest, sourced, mildly conservative assumption buys more credibility than a tab of aggressive ones, because reviewers extrapolate: however you built this number is how you built all of them.
3. Revenue Built from Drivers, Not Wishes
Revenue should compute from things you can count: leads times conversion times average contract value, units times price, billable hours times realized rate, customers times retention times expansion. A flat "40 percent annual growth" line is not a forecast; it is a hope with formatting. Driver-based revenue does three jobs at once: it makes the forecast testable (reviewers can push on each driver separately), it connects the model to operations (a pipeline miss shows up as a revenue miss mechanically), and it proves you understand your own growth engine, which is half of what the reviewer is actually evaluating.
4. Costs That Scale the Way Reality Scales
Split costs into fixed and variable, and make the variable ones move with their drivers: fulfillment with orders, support headcount with customers, payment fees with revenue. Then check the tells reviewers check: headcount that magically stays flat while revenue triples, gross margin that improves every single year without a stated mechanism, and marketing spend that falls as a percentage of revenue while growth accelerates. Real operating leverage exists, but it has causes, automation, pricing power, capacity utilization, and a credible model names them.
5. Unit Economics That Survive Arithmetic
Whatever your business, there is a unit, an order, a customer, a project, and the model should show its economics explicitly: what one unit costs to acquire and serve, what it returns, and how fast it pays back. For recurring-revenue businesses that means CAC, lifetime value, and payback months; for project businesses, realized margin per engagement; for commerce, contribution margin per order after fees, fulfillment, and returns. Reviewers compute these themselves from your numbers if you do not present them, and it reads far better when you have done the math first, including the unflattering parts.
6. Cash Runway and a Downside Case You Volunteer
The three-statement model must produce a cash line, and the cash line must be tested: what happens at 75 or 80 percent of planned revenue, with collections arriving 15 days slower? Presenting that scenario unprompted, with the survival math and the actions you would take, is the single highest-credibility move available in a model review, because it demonstrates the thing every reviewer is really probing for: that you understand profit and cash are different resources with different failure modes. The mechanics of that distinction are the subject of our guide to profit, cash flow, and ROI.
7. A Forecast That Honors Your History
The forecast's first year should reconcile against your trailing actuals, same margins, same seasonality, same cost ratios, or explain precisely why not. The classic failure is the hockey stick that begins the month after the raise: 20 percent historical growth becoming 200 percent projected, with no new mechanism named. Reviewers do not require modest projections; they require causal ones. "Growth doubles because we are hiring three salespeople with quota history at this deal size" is a claim that can be tested. "Growth doubles" alone is a claim about hope.
What Reviewers Check, at a Glance
| Check | What good looks like | The red flag |
|---|---|---|
| Structure | Inputs / calculations / outputs separated; numbers traceable | Hardcoded constants inside formulas |
| Assumptions | Labeled, sourced, testable | Uniformly optimistic; "industry standard" with no source |
| Revenue | Built from countable drivers | Flat growth percentage |
| Costs | Fixed/variable split; scaling with named causes | Margins improve annually by magic |
| Unit economics | Presented per unit, including payback | Only company-level averages shown |
| Cash | Downside scenario volunteered with survival math | Single optimistic case; runway untested |
| History | Year one reconciles to actuals or explains the break | Hockey stick starting the month after funding |
What We See in Practice
Three patterns from model reviews we have run and repaired. First, most models fail on structure before anyone reaches the assumptions: the reviewer gets lost in twenty minutes, stops, and the meeting becomes about the model instead of the business. Second, the question owners least expect and most often stumble on is the simplest: "walk me from this quarter's actual revenue to next quarter's projection." Founders and owners who can answer in drivers, this pipeline, that conversion, this capacity, clear the bar almost regardless of the spreadsheet's polish. Third, the downside case remains the great differentiator; in a capital environment that stayed selective into 2026, we have watched it change the tone of reviews more than any growth number, and it costs an afternoon to build. A model that clears all seven checks is also most of the way to the diligence package described in making your numbers investor-ready.
Frequently Asked Questions
What do investors look for in a financial model?
Seven things, roughly in order: navigable structure, labeled and sourced assumptions, revenue built from countable drivers, costs that scale believably, explicit unit economics, cash runway tested under a downside case, and a forecast that reconciles with historical results. The model is read as evidence of how the operator thinks, not just what they project.
How many years should a financial model project?
Three years of monthly or quarterly detail is the working standard for private businesses; five-year precision at SMB scale is theater, and reviewers know it. Year one should be tight and reconcile to current actuals, year two driver-based, year three directional. Depth of mechanism matters far more than length of horizon.
What is a driver-based financial model?
A model where revenue and costs compute from countable operational inputs, leads times conversion times deal size, orders times average order value, hours times rate, instead of flat growth percentages. It is the format sophisticated reviewers expect because it makes assumptions testable and ties the forecast to how the business actually grows.
What are the biggest red flags investors see in financial models?
Hardcoded numbers buried in formulas, uniformly optimistic assumptions, hockey-stick growth with no named mechanism, margins that improve every year by magic, missing downside scenarios, and forecasts that ignore the company's own historical performance. Any two of these together usually end serious interest, whatever the business's real merits.
Should I hire someone to build my financial model?
If a raise, loan, or sale is approaching and your current model would fail the checks above, yes, and the right builder works with you rather than for you, because you will defend every assumption in the room yourself. A fractional CFO or FP&A engagement typically builds the model and the fluency together; the model alone is half the product.
The Bottom Line
A financial model persuades when it is legible, causal, and honest: structure a reviewer can navigate, revenue built from drivers, costs that scale for stated reasons, unit math done in the open, a downside case you volunteer, and a forecast your own history would recognize. Build that and the model conversation gets short, which is exactly what you want; the meeting should be about the business.
Building models that clear this checklist is core work in our FP&A service. If capital is on your horizon, talk to us and we will review your current model against the seven checks, honestly.
