Every venture-stage founder has a financial model. Most of them are not investor-ready. The difference is rarely the numbers. It is the structure underneath them.
A deck spreadsheet exists to make the deck math out. An investor-ready model exists to survive diligence. Different objects, different bars.
The seven structural elements.
1. Driver-based logic, not hard-coded outputs.
Revenue cannot live as a typed number. It has to be the output of a calculation: customers times price, or users times conversion times average revenue. When an investor changes one assumption, every downstream cell should recompute. Hard-coded outputs are the single fastest way to lose credibility in a model walkthrough.
2. Centralized assumptions tab.
Every input the model uses (pricing, churn, growth, COGS, hiring, capital structure) lives in one tab. Color-coded blue for inputs, black for calculations. This is convention. Violating it tells an analyst you have never built a model that has gone through diligence.
3. A real three-statement structure.
Income statement, cash flow statement (indirect method), balance sheet. Tied together. Ending cash on the cash flow ties to the cash balance on the balance sheet. The balance sheet balances. Working capital movements (AR, AP, deferred revenue) flow correctly between statements. Most founder-built models stop at the income statement. Investors notice.
4. Monthly resolution for at least 24 months.
Annual numbers hide everything. Hiring ramp, seasonality, working capital swings, cash dips. Investors model monthly because the monthly cash trajectory is what kills companies. If your model is annual, an analyst will rebuild it monthly to evaluate it, and they will not enjoy the exercise.
5. Unit economics, computed not asserted.
CAC, LTV, payback period, contribution margin. These cannot be typed-in numbers. They have to be computed from the underlying customer cohorts, revenue lines, and acquisition spend. When an investor asks "how did you get to a six-month payback," the answer should be in the model, not in the founder's head.
6. Scenario logic, not multiple files.
Base case, downside, upside. Toggleable inside the model, not three separate spreadsheets. Scenarios are about the same structure under different assumptions. If a founder hands me three separate files, I know none of them are reconciled to each other.
7. A use-of-funds tab that ties to the model.
When a founder says they are raising $2 million, the model should show what $2 million buys, when it gets deployed, and what runway it produces. The use of funds is not a slide. It is a tab inside the model with hiring sequencing, marketing spend, and infrastructure costs all tied to the operating plan.
The bars investors carry in their heads.
The analyst is not only checking whether the model works. They are checking its outputs against a short list of numbers they already know. For a software company at seed or Series A, the rough bars are these.
- Gross margin of 70% or better. Under 60% and the questions turn to whether this is software at all.
- Burn multiple, net burn divided by net new ARR, under 2x. Under 1.5x reads as efficient. Over 3x needs a story.
- CAC payback under 18 months at seed, trending toward 12 by Series A.
- Logo churn under 2% a month for SMB customers, under 1% for mid-market and enterprise.
- Runway of 18 to 24 months after the round closes, with the month of lowest cash visible on the page.
Your model does not need to hit every bar. It needs to compute each one honestly, and the deck needs to say the same number the model does.
What an analyst actually does with it.
When a Series A analyst gets your model, they do three things in the first hour. They flex revenue down 30%. They extend the hiring plan by a quarter. They check whether the balance sheet still balances when they change the working capital assumption.
If any of those three tests breaks the model, the answer to "would you like to invest" gets harder to defend internally. If all three hold, the analyst moves to the next question: is the underlying business worth funding.
A driver chain, with numbers.
Here is what driver-based means in one line of revenue. Start with 100 customers on a $249 monthly plan. Add 20 new customers a month from a defined acquisition budget. Lose 2% a month to churn. Month one ends with 118 customers and $29,382 of monthly revenue. Month twelve ends with about 294 customers and $73,100. Change the churn assumption to 4% and month twelve lands near 255 customers and $63,500, and every cash balance below it moves with it.
That is the whole test. One assumption, in one cell, and the entire model reacts. If an investor has to ask which cells to change, the model has already failed. The 24-month forecasting framework covers what goes in the first three tabs.
The model is not the pitch.
Many founders treat the financial model as a downstream artifact of the pitch deck. Build the deck, then back into the model to support the numbers. This is backwards.
The model is the system. The deck is a visual summary of the system. Build the model first, let it teach you what the business actually looks like, and then build the deck off the model's outputs. This sounds slower. It is dramatically faster.
How I build one in thirty days.
- Week one: read. Bank, ledger, pricing page, pipeline, payroll. The model starts from what the company actually spent last month, not from a template.
- Week two: drivers. Customers, price, churn, hiring dates, cost per unit. Every input on one tab, in blue. Nothing else is typed.
- Week three: statements. Income statement, cash flow, balance sheet, monthly, 36 months out. Ending cash on the cash flow equals cash on the balance sheet every month, or the model does not ship.
- Week four: stress. Revenue down 30%, hiring slipped a quarter, collections slowed 15 days. If the lowest cash month moves more than the founder expects, we fix the plan before an investor finds it.
The Runway and Burn Modeler is the smallest version of week four. Put in your current burn and planned hires and it shows the month of lowest cash and the date fundraising has to start.
One question to ask your model.
If you changed the customer growth rate by one percentage point, would your model produce a different cash balance in month twenty-four? If yes, you have a model. If no, you have a spreadsheet.