How to Start a Business Plan: A Founder's Step-by-Step Guide

I built Preuve AI a few months ago. Founders have run more than 4,500 startup ideas through it since launch, and the patterns from those scans are what shaped this guide. The single most common mistake I see when founders start a business plan is starting at the wrong place: a blank Word document instead of a customer conversation.

Starting a business plan is a research project before it is a writing project. When that order is right, the writing itself takes 60 to 90 minutes. Get the order wrong and the same draft eats two weeks of polish on a document that does not reflect a real market.

This guide walks through how I would start a business plan today, in this order: validate the problem, pick the right format, draft the seven core sections, and stress-test the financials. Use it as a checklist if this is your first plan. Use it as a debug tool if your last one did not get traction.

Why most founders start a business plan the wrong way

Founders treat the business plan as the first deliverable. They open a template, fill in the executive summary, invent a five-year revenue projection, and call themselves prepared.

The U.S. Bureau of Labor Statistics reports that about 20% of new businesses fail within the first year, and 50% within five years. The plan was rarely the cause. The lack of validation behind the plan almost always was.

A 2010 study from Babson College, replicated several times since, found that founders who wrote a plan before talking to customers were no more likely to launch than those who did neither. Founders who validated first and then planned were 2.6 times more likely to reach revenue. The plan is a translation layer for evidence. Without evidence, the document is creative writing.

The way I think about it: the business plan is a hypothesis test, not a sales document. If you cannot name a real customer, a real price, and a real channel for the first sale, the next thing you should be doing is not opening a template. It is calling people.

For the full case on when to write versus when to wait, I wrote a separate piece on the startup business plan trap. This guide assumes you have already decided you want to write one and need to know how to start.

Step 1: Validate the problem before you write

Validation is not optional. Every section of a business plan that works traces back to evidence collected before the document existed.

The five questions every business plan secretly answers

Before opening a template, answer these in writing. If you cannot, the plan is not ready to start.

1. Who specifically has this problem? Job title, industry, company size, geography. "Small business owners" is not specific. "Solo dental practice owners in U.S. cities of 50,000+ population" is specific.
2. How are they solving it today? Name the tool, the workaround, the consultant, or the spreadsheet. If they are not solving it, the problem is probably not painful enough to fund a solution.
3. What does the problem cost them today? In hours, dollars, customers lost, or revenue forgone. A problem without a cost is a curiosity, not a market.
4. Will they pay to solve it? Not "would they". Will they. The only honest answer comes from a price quote and a "yes" or "no" in a real conversation.
5. How do you reach them? A specific channel: a subreddit, a trade publication, a LinkedIn group, a conference, a referral source. "Online marketing" is not a channel.

If you cannot answer all five with specifics drawn from real conversations, the business plan you write next will be fiction. The plans I have read where the founder skipped this step share a tell: the language drifts to abstractions, the numbers are always round (it is $10M, never $7.4M), and the customer never has a face.

A 60-minute validation framework

Before you write a business plan, run this. It costs nothing and saves weeks.

• 15 customer interviews, 20 minutes each. Open-ended questions, no pitching. Capture their exact words.
• 3 competitor purchases. Buy what your future customer buys today. Note what they do well and where the pain remains.
• 1 landing page test. Describe the solution, ask for an email, run $50 of traffic. Conversion rate is your first demand signal.
• 1 willingness-to-pay test. Quote a price to five interview subjects. Count the "yes" answers.

I built Preuve AI to compress the desk-research portion of this work into 90 seconds, but the customer conversations cannot be skipped. No tool replaces a 20-minute call with a real buyer.

Step 2: Pick the right business plan format

Three formats cover 95% of cases. Picking the right one is the second decision after validation, and getting it wrong wastes time. Start by asking who reads the document and what decision they need to make.

If you cannot name a specific person who will read 25 pages, do not write 25 pages. The longest plan is not the most credible plan. The plan most aligned with the reader's decision is.

When the Business Model Canvas beats both

Alex Osterwalder's Business Model Canvas (Strategyzer, 2010) is nine boxes on one page: customer segments, value propositions, channels, customer relationships, revenue streams, key resources, key activities, key partners, cost structure. It is not a business plan. It is a thinking tool that produces the inputs for one.

Run the canvas first. Fill it in pencil. Refill it after every five customer interviews. When the canvas stops changing week to week, you have enough signal to write the plan. Most founders should spend two to four weeks on the canvas before opening a longer template.

Step 3: Draft the seven core sections (in the right order)

Every business plan that works contains these seven sections. The U.S. Small Business Administration's official traditional template uses the same structure with two optional additions (funding request and appendix).

Counter-intuition: do not write them in this order. Write them in the order shown in the "draft order" column. The executive summary is last because it summarizes what you wrote, not what you intend to write.

Section 3 deep-dive: market analysis without guessing

The market analysis is the section that sinks the most plans. Founders cite a $400 billion industry and call it done. A reader notices in 30 seconds.

Three numbers, in this order, with sources:

• TAM (Total Addressable Market): the entire revenue opportunity if you captured 100% of the space. Source: government data (U.S. Census, Bureau of Economic Analysis), industry reports (IBISWorld, Gartner, Statista), or trade associations.
• SAM (Serviceable Addressable Market): the slice you could realistically reach once you account for geography and language constraints, plus the channels you can actually sell through today. Usually 5-20% of TAM.
• SOM (Serviceable Obtainable Market): what you can actually win in the first three years. Bottom-up calculation: customers reachable × conversion rate × average revenue per user.

I wrote a full guide on calculating TAM SAM SOM with real sources if you need the formulas. The shortcut: if your TAM is more than $50 billion and your SOM is more than $10 million in year one, your numbers are wrong.

Section 7 deep-dive: financial projections that hold up

The financial section needs four artifacts at minimum:

1. Three-year profit and loss (revenue, cost of goods sold, gross margin, operating expenses, EBITDA).
2. 18-month cash flow plan (monthly granularity for the first 18 months, then quarterly).
3. Break-even analysis (units sold or revenue required to cover fixed costs).
4. Assumption appendix (every number traces to a stated assumption).

Skip the five-year forecast. The further out you project, the more you signal to the reader that you are guessing. Investors mostly care about the next 18 months, lenders about the next 36, and nobody I have met takes year five seriously regardless of who built the model.

Anchor every assumption in a comparable. Pricing should reference at least two real competitor prices. Customer acquisition cost should reference a benchmark from a similar business model (the OpenView SaaS Benchmarks Report publishes free data each year). If a number cannot be benchmarked, label it as a hypothesis to test in the first 90 days.

Step 4: Start your first draft in 60 minutes

Once validation is done and the format is picked, the first draft is fast. Here is the 60-minute first-pass I run when starting a business plan from scratch.

1. Minute 0-10: brain-dump the market analysis. Three numbers (TAM, SAM, SOM), three competitors, one customer quote. Bullet form. No prose yet.
2. Minute 10-20: brain-dump the product section. One sentence on what it is. Three sentences on how it works. One sentence on what makes it defensible.
3. Minute 20-30: brain-dump marketing and sales. One channel you have already tested. One pricing model with a real comparable. One sales process step.
4. Minute 30-45: brain-dump financials. Year-one revenue (units × price), year-one costs (fixed + variable), break-even month. No five-year forecast yet.
5. Minute 45-55: brain-dump company and team. Mission in one sentence. Founders' relevant background in three sentences. Open roles in a list.
6. Minute 55-60: write the executive summary last. One paragraph that compresses every bullet above. If it does not flow, the underlying sections are weak. Fix them, not the summary.

Print the result and read it as if a stranger handed it to you. Three flags to watch for: vague nouns ("solutions", "synergies"), unsourced numbers, and adjectives doing the work that should belong to nouns. Strike them. What is left, even if it looks thin, is the actual skeleton of a real plan, and that is more than most founders have at this stage.

Free business plan templates worth using

Skip paid template marketplaces. The free ones from credible sources are better, and the credibility transfers when a lender or investor recognizes the format.

Other free templates worth using:

• U.S. Small Business Administration: two free templates (traditional and lean startup) at sba.gov. The default for any plan submitted to a U.S. lender.
• SCORE: 14 industry-specific templates (food service, retail, professional services, manufacturing) at score.org. Backed by SBA mentors.
• Strategyzer: the original Business Model Canvas in PDF and digital, free at strategyzer.com.
• Leanstack: Lean Canvas (Ash Maurya's startup-focused variant of BMC), free at leanstack.com.

For Google Docs versions, both SBA templates are publishable to Google Docs in two clicks. Use the SBA file directly. Do not pay $29 for a Notion template that is the same content with a different cover page.

Seven beginner mistakes that sink first business plans

I have read hundreds of first plans. The same seven mistakes appear in roughly that order:

1. Writing the executive summary first. You compress assumptions, not facts. Always last.
2. Citing TAM only. $50 billion TAM with no SOM is a red flag. Lenders and investors look for the SOM first.
3. Five-year revenue projections with no Q4 plan. If you cannot map the next 12 weeks in detail, year five is fiction.
4. Using "we believe" or "we anticipate" instead of citing data. Replace every "we believe" with a number and a source. If you cannot, drop the claim.
5. Listing 12 competitors, none of them real substitutes. Three real competitors beat 12 vague ones. Name the actual product the customer uses today.
6. Generic customer descriptions. "Small business owners" is not a customer. "Independent dental practice owners in U.S. cities of 50,000+ with 1-3 employees" is.
7. No pricing comparable. If your $99/month price is not anchored to two competitors' published prices, the financial projections are guesses.

The pattern: most first-plan mistakes come from filling in a template before doing the underlying research. The plan is downstream of the research, not a substitute for it.

When to stop planning and start selling

The business plan is a checkpoint, not a finish line. The moment to stop polishing the plan and start selling is the moment three conditions are true:

• You can describe the customer, the problem, and the price in one sentence each, without notes.
• You have at least three customer interviews on record where the buyer used the words "I would pay for that" with a specific dollar number attached.
• You have a 12-week plan to make the first sale, and you can name the channel.

If those three are true, the plan is already good enough. Close the doc and go sell. Whatever revision you do six months from now will be sharper because the first ten customers will have rewritten parts of it for you, often without realizing they did it.

Frequently asked questions

The shortest path forward

Starting a business plan is a sequence, not a draft. The work goes in this order: validate first, then pick the format that matches whoever is going to read the document, then draft the seven sections in evidence order with the executive summary written last, then stress-test the financials against published benchmarks. The point at which it is "good enough" is when you can sell from it, not when it is perfect.

If you are at step zero (idea in head, no customer conversations yet), do not open a template. Start with the validation work. I built Preuve AI for that step: it compresses the desk-research portion (TAM, competitors, demand signals, market sources) into 90 seconds so you can spend the saved hours on customer interviews. The plan you write afterwards ends up faster to draft and noticeably more defensible, partly because it is shorter and partly because the evidence is already in front of you.

Two more guides if you want to keep going: how I validate startup ideas before any plan exists, and the customer discovery questions I use in every interview.