How to Create a Startup Business Plan
What a plan has to do
Every section exists to prove one claim. Write each section by asking what a sceptical reader would need in order to believe that claim, then supply exactly that — and mark clearly where you are estimating rather than reporting. A stated estimate is credible. An unstated one, discovered by the reader, costs you the whole document.
Section by section
One-page summary
Proves: That the business can be understood in ninety seconds.
- What the business does, in one sentence a non-expert understands.
- Who the customer is and what it currently costs them not to solve this.
- The stage you are at and the evidence behind that claim.
- What you are asking for, if anything, and what it buys.
Problem and customer
Proves: That a specific group of people has a costly problem — established independently of your solution.
- The problem stated in the customer's language, not yours.
- A narrow, reachable first segment, not a demographic.
- Evidence: conversations, workarounds observed, money already being spent on alternatives.
Solution and why now
Proves: That your approach fits the problem, and that something has changed to make it viable today.
- How the customer's workflow differs after you exist.
- What is genuinely hard about doing this, and why that is durable.
- The change — technology, regulation, cost, behaviour — that opened the window.
Market and competition
Proves: That the market is large enough to matter and that you know who you displace.
- A market size built bottom-up from customers × price, not a headline industry figure.
- Named competitors, including the status quo: a spreadsheet, an agency, or doing nothing.
- Why a customer switches from that specific alternative to you.
Business model and unit economics
Proves: That an individual sale makes money before scale is considered.
- Price, variable cost per unit, and contribution margin.
- Cost to acquire a customer, and how you measured or estimated it.
- Break-even volume, and how far current volume is from it.
- Retention or repeat rate, if the model depends on it.
Go-to-market
Proves: That you can reach customers repeatably, at a cost the price supports.
- One or two channels you have actually tested, not five you might try.
- The cost and conversion rate you observed, with sample size stated.
- What the first hundred customers look like and where they come from.
Operating plan and team
Proves: That this team can execute this plan on this timeline.
- The next three to four milestones, each with a date and a definition of done.
- Who does the work, and which capability gap you know you have.
- What each hire unlocks, expressed as a milestone rather than a headcount.
Financials and assumptions
Proves: That the numbers were derived, not invented.
- Monthly cash view for 12-18 months: revenue collected, costs, cash balance.
- Every driver listed separately, with its source marked as measured, benchmarked, or estimated.
- Runway, break-even point, and what has to be true for both.
- A downside case where the main growth assumption is halved.
Risks
Proves: That you have looked at the failure modes deliberately.
- The three risks that would end the business, named plainly.
- The leading indicator you watch for each one.
- What you would do if it materialised — decided in advance.
The numbers to get right
Four figures carry most of a plan's credibility, and all four are calculable rather than argued:
Price minus variable cost per unit. If this is negative, no volume saves the business, and a reader will spot it in seconds.
Fixed costs divided by contribution margin. State it in units and in revenue, and say how far today's volume is from it.
Cash divided by net burn. This sets the deadline on every milestone in your operating plan — they must fit inside it.
Measured if you have run a channel, clearly labelled as an estimate if you have not. Never quietly assumed.
Two of these can be computed right now: break-even and runway. Put the outputs directly into the financial section with the inputs shown beside them.
What gets a plan rejected
Revenue that triples in year three without a named mechanism reads as a hope. Tie every increase to a channel, a hire, or a price change.
"1% of a $50B market" is not a plan. Build market size from customers you can name and reach, multiplied by a price you have tested.
There is always an alternative, even if it is a spreadsheet. Claiming none signals you have not looked.
An unsourced conversion rate that drives the whole model is the fastest way to lose a reader's trust in everything else.
Do this next
- Write the one-page summary first. If it is hard, the plan underneath is not settled yet.
- Calculate break-even and runway before writing the financial section, not after.
- Mark every driver in your model as measured, benchmarked, or estimated.
- Halve your main growth assumption and check whether the business still survives.
- Confirm the plan matches the stage you are actually at — the startup process roadmap gives the exit condition for each one.
Common questions
How long should a startup business plan be?
For most purposes, 5 to 12 pages plus a financial model. Length is not the signal — a reader is checking whether your assumptions are stated, sourced, and internally consistent. Long plans usually hide thin evidence.
What sections does a startup business plan need?
Problem and customer, solution, market and competition, business model and unit economics, go-to-market, operating plan, team, financials with stated assumptions, and risks with mitigations. A one-page summary sits in front.
What do investors and lenders look at first?
Usually the team, the unit economics, and the assumptions behind the revenue line — in that order. They are testing whether the numbers were derived or invented, which is why every projection should trace back to a named assumption.
Keep going
Compute the number your financial section needs first.
The evidence that turns plan assumptions into claims you can defend.
Who the plan is for changes what it must prove.
The deadline your operating plan has to fit inside.