How to write a business plan
AI has changed how business plans get written. This guide doesn't recite theory at you: it gives you ready-to-use prompts, mini-tools that do the maths and a method for testing your plan against reality.
Updated July 2026.
Free, no credit card required.
Three resources, zero filler
Each one is built to be used, not just read. Start with whichever you're missing today.
AI prompt library
Twelve professional prompts (market research, business model, assumptions, pitch), ready to paste into the AI of your choice.
Break-even calculator
Your numbers, a live calculation, one clear answer: how much do you need to sell to stop losing money?
Business models by sector
How money comes in across your industry: the models that work, their traps, and where to start.
- Nearly 1 in 3
- French companies founded in 2018 were gone before their fifth birthday
- Insee, 2025
A business plan doesn't run itself.
Writing the plan is the start. Testing it against reality every month is what makes it useful.
The 5 steps to a business plan that holds up
There's no secret to the method. What changes everything is following it all the way through, especially after launch.
-
Nail down your business model
Before you build a single table, answer four questions: who has the problem, what are you selling, who pays, and how the money comes in (per unit, by subscription, on commission). If the answer doesn't fit in three sentences, the rest of the plan will only dress up the fog.
With BeretPlan: start from your industry's template or a generic one, or import your existing budget, instead of a blank page.
Explore business models by sector -
Put numbers on your assumptions
A business plan isn't a prediction: it's a list of assumptions you're willing to own. A price, a volume, a conversion rate, your costs. Write them down, one by one: they are what your banker will challenge, not your layout.
With BeretPlan: every assumption becomes a named variable. Change it once and the whole plan updates.
Prepare your assumptions with the AI promptsVariables tab of a BeretPlan plan: assumptions grouped (acquisition, service revenue, subscription revenue with price and conversion rate per tier). Service revenue is the monthly acquisition times the service share times the average service order; picking another average service order recalculates the service revenue of every month, January to June. Configuration VariablesFilter variables...Configure...Acquisition
- Monthly acquisition
- 300
- Quarterly growth
- 20%
Service revenue
- Average service order
-
- Service share
- 1%
Subscription revenue
Price Conversion Pro €29 10% Max €99 2% Service revenue · €
- Apr 3,200
- May 3,200
- Jun 3,200
Illustrative example · Try it: change the average service order
-
Build the forecast
Sales, costs and profit, month by month, over two to three years. The goal isn't precision (nobody can predict month 27), it's consistency: revenue must flow from your assumptions, never the other way round. First thing to check: the month you stop losing money.
With BeretPlan: the income statement and the dashboards are generated from your plan, with formulas you can actually read, and no #REF!.
Calculate your break-even pointSaaS forecast in BeretPlan: the Subscriptions line is Pro users times the Pro price plus Max users times the Max price, the Services line is service orders times the average order, month by month from January to June 2027, and the Revenue total follows. Selecting a revenue line reveals its formula in descriptive mode and its inputs, like the app's configuration panel. Forecast · SaaS startup
€ · monthly · 2027
Line Type Trend Jan Feb Mar Revenue 3,864 5,328 6,792 1,464 2,928 4,392 2,400 2,400 2,400 Name
SubscriptionsServices
Value
Pro users×Pro price+Max users×Max price
Inputs
Pro price = €29 Max price = €99 Pro users Max usersService orders×Average service order
Inputs
Average service order = €800 Service ordersIllustrative example · Try it: select a revenue line
-
Test the plan against reality
The step almost everyone skips, and the one that turns a document into a tool. Every month, enter what actually happened and look the variances in the eye. A variance isn't a mistake: it's information, and it's worth a lot.
With BeretPlan: actuals are entered line by line against the forecast, and variances are coloured by their real impact. An expense drifting off course stands out, even while the curve is going up.
Example of the actuals grid in BeretPlan: actuals entered month by month against the forecast, each variance coloured by its impact. Tap a cell for the actual, the plan and the variance. An expense above plan shows in red, revenue above plan in green. Actuals · SaaS
Line Type Trend Mar 27 Apr 27 May 27 Jun 27 Revenue €6,432 ▼ €8,696 ▼ €10,902 ▼ €11,322 ▼ Subscriptions Services Expenses €17,540 ▼ €18,030 ▲ €23,120 ▲ €23,240 ▲ Salaries Marketing Cloud Servers and cloud Illustrative example · Try it: tap a cell
-
Iterate
Reality is always right. When a price turns out too low or a channel doesn't deliver, update the assumption, not the story around it. A living business plan gets rewritten; that's precisely how you know it's earning its keep.
With BeretPlan: change one variable and the forecast, the income statement and the dashboards all recalculate. No more “final_v3_REAL.xlsx”. And with alternative scenarios, compare several versions of your plan (“what if I hired earlier?”) without duplicating it.
Example of a projection in BeretPlan, for a SaaS startup: actuals run January to June, then Pro users and subscriptions are projected along the trend to December. Projected months are italic with a dotted underline, with their variance to plan: the year lands below plan. The crystal-ball button shows or hides the projection. Projection · SaaS startup
Line Jun Jul Aug Pro users 164 −17% 189 241 −22% 214 284 −25% Subscriptions €8,122 −15% €9,342 €11,741 −20% €10,562 €13,879 −24% Illustrative example · Try it: show or hide the projection
A plan that holds up is a plan you steer.
Set your assumptions and get a forecast that recalculates itself: a complete budget in under 2 hours.
Free, no credit card required