Budgeting, Forecasting & Building a Simple Financial Model

So far, most of business finance has been about looking back — what did we earn, what did we spend, what is the cash today. This section flips you around to look forward. You are going to learn how to make a plan for the future in numbers, check the plan against reality each month, and fix the plan when reality disagrees.

Do not be scared. A "financial model" sounds like something only a banker in a suit can build. It is really just a spreadsheet that says: "If these few things happen, here is the money that comes out." If you can multiply and add, you can build one.

Three words people mix up: budget, forecast, and model

These three words get used as if they mean the same thing. They do not. Let me define each in plain English.

• Budget — your plan for a period (usually one year). It is a promise to yourself: "We plan to make $40,000 a month and spend $30,000 a month." You set it once, usually at the start of the year, and you mostly leave it fixed so you can measure yourself against it.
• Forecast — your best honest guess of where you are actually heading, updated as new facts arrive. The budget says "here is what we planned." The forecast says "here is where we are really going, based on what we now know."
• Financial model — the machine (a spreadsheet) that produces both. You feed it assumptions (price, number of customers, costs), and it calculates revenue, profit, and cash. Change one assumption and every number downstream updates by itself.

Two ways to build a revenue plan: bottoms-up vs top-down

Before any model, you need a number for revenue (money coming in from sales). There are two directions to reach it. Smart founders use both and compare them.

Top-down: start big, shrink to your slice

You start with the total market and take a small share of it. TAM means Total Addressable Market — the total money everyone could spend on your kind of product.

Bottoms-up: start with what you actually do

You build revenue from your real activities: how many customers, at what price, how often. This is slower but far more honest, because every number is something you can do something about.

Two common bottoms-up formulas:

Quick definitions: Traffic = visitors to your site. Conversion rate = the % of visitors who actually buy. AOV (Average Order Value) = the average dollars per order.

Driver-based modeling: make assumptions explicit and editable

A driver is an input number that "drives" the result — like price, customer count, or churn. An assumption is your guess for that driver before you have real data.

The golden rule: never type a final answer directly into a cell. Type the drivers, and let the spreadsheet calculate the answer. This is "driver-based modeling." If you write revenue as "$100,000" by hand, you've learned nothing. If you write traffic × conversion × AOV, you can change any driver and instantly see the effect.

Three scenarios: base, best, and worst case

The future is unknown, so don't pretend you know it exactly. Build three versions by changing your key drivers:

• Base case — what you genuinely expect (your honest middle bet).
• Best case — things go well (higher conversion, faster growth).
• Worst case — things go badly (slow sales, higher costs). This one matters most — it tells you when you'd run out of cash.

Rolling forecast vs annual budget

An annual budget is set once a year and frozen, so you can measure yourself against it. A rolling forecast is updated every month and always looks the same distance ahead — as one month finishes, you add a new month at the far end, so you always see roughly 12 months forward.

You don't have to pick one. The strongest setup keeps both: the frozen budget gives you accountability (did we hit the plan?), and the rolling forecast gives you agility (where are we really heading now?).

Variance analysis: plan vs actual, then act

Variance just means the difference between what you planned and what really happened. Variance analysis is the monthly habit of putting plan and actual side by side and asking why they differ — then doing something about it.

A simple month-by-month model layout

Here is a clean layout. Months across the top, line items down the side. Every revenue number is calculated from the drivers above it — nothing is typed in by hand except the assumptions.

How the flow works, top to bottom:

  Drivers (you type these)
  traffic  conversion  AOV  costs
        \      |       /     |
         v     v      v      |
        REVENUE = T x CR x AOV
                  |          |
                  v          v
              REVENUE  -  COSTS  = PROFIT
                                    |
                                    v
            CASH(last month) + PROFIT = CASH(now)

Best practices that separate good models from junk

• One assumptions tab. Keep every driver (price, conversion, churn, salaries) in one place, clearly labeled. The rest of the sheet only does math on those cells. To test a new idea, you change one tab — not 200 scattered cells.
• Don't fake precision. Writing "$1,284,447.18" for next year is silly — you don't know it to the penny. Round numbers ($1.3M) are more honest and easier to trust.
• Update monthly. The best model is the one you actually keep current. A simple sheet you refresh every month beats a fancy one nobody has touched in three months.
• Color-code inputs. Make typed-in driver cells one color (say blue) and calculated cells another. Then you instantly see what's a guess and what's math.

Sources: Forecastio, Finro, 8020 Consulting, Onetribe Advisory, PrometAI, Inflection CFO, Polaris Growth.