How to Write a Startup Idea Worth Validating: 7 Formulas That Beat Brainstorming

The single biggest reason startup ideas fail before they've even been tested isn't competition, market timing, or funding. It's the idea itself — described too vaguely to validate, too emotionally to critique, or too big to know where to start. The idea is the bottleneck, and most founders treat it as the easy part.

It's not. The ability to write a startup idea as something that can be tested is a learnable skill, and it is the highest-leverage activity in the entire pre-build phase. A well-written idea saves you 30 hours of validation work. A poorly written idea can fool you into spending six months building the wrong thing.

This post is a practical, opinionated guide. No philosophical nonsense about "finding your passion." Just seven formulas that work, the failure modes each one fixes, and a worked example of how to take a vague notion and turn it into a one-paragraph hypothesis you can actually test.

Why "I want to build a SaaS for X" is not an idea

The most common mistake is confusing a category with an idea. "I want to build a SaaS for plumbers" is a category, not an idea. It has no job-to-be-done, no specific buyer, no specific moment of pain, and no hypothesis. You can validate a category the way you can validate "the weather" — it has no falsifiable claim, so any answer is true.

A real startup idea has three properties:

1. A specific buyer. Not "small businesses" or "developers" or "creators" — a named buyer with a named context. "Independent residential HVAC companies in the US with 5–50 technicians" is a specific buyer. "Small businesses" is not.
2. A specific job-to-be-done. What is the buyer trying to accomplish, in what moment, and what's painful about the current way they accomplish it? "Schedule service calls faster" is too vague. "Reduce the time between a customer's call and a confirmed appointment from 4 hours to 15 minutes" is a specific job.
3. A falsifiable hypothesis. "I believe I can charge $200/month for this and that at least 200 HVAC companies will pay because the current tools take 8 hours/week of admin time." That can be proven wrong. "I think plumbers would like a better tool" cannot.

The seven formulas below are different ways to fill in those three boxes. None of them is a complete substitute for thinking, but they all force specificity, which is the thing most "ideas" lack.

Formula 1: The Verb + The Object + The Constraint

I help [specific buyer] do [specific verb] to [specific object] in less time / for less money / with less pain than they can today.

Example: "I help independent US tax preparers file partnership returns (1065s) in under 30 minutes, instead of the 6–10 hours they currently spend."

Why it works: the verb forces a job; the object forces a domain; the constraint forces a measurable improvement. If you can't fill in all three, you don't have an idea yet.

Failure mode: "I help SMBs do business faster." No verb, no specific object, no constraint. Try again.

Formula 2: The Painful Workflow

Pick a workflow you have personally watched someone do, that you found painful to watch. Describe it step by step, and identify the single step that wastes the most time or causes the most errors.

Right now, [buyer] has to [painful step] in order to [accomplish goal]. This takes [X hours / days / clicks] and fails [Y% of the time]. I can reduce it to [X'] with [approach].

Example: "Right now, an insurance broker has to manually re-key every client policy into three different carrier portals. This takes 4 hours per client and has a 12% typo rate. I can reduce it to 20 minutes per client with a single API integration that pre-fills carrier forms from the broker's existing CRM."

Why it works: you've anchored the idea in observed reality, not opinion. The numbers (hours, error rate) are claims you can go test with 5 buyer interviews.

Failure mode: making up the numbers. If you haven't actually watched the workflow, the numbers are fiction. Go watch the workflow.

Formula 3: The Two Lists

Take an existing product the buyer uses for [job A]. Identify 5 things it does well. Identify 5 things it does badly. Pick the three worst "bad" items and design a new product that fixes only those three.

Example: "QuickBooks does payroll, invoicing, and tax filing well; it does approvals, multi-entity, and audit-trail badly for mid-market customers. I'll build a tool that does only approvals, multi-entity, and audit-trail, and integrates with QuickBooks for the rest."

Why it works: you're starting from a real, paying user's frustrations, not from a hypothesis about a hypothetical user. The "two lists" approach also produces a sharp positioning: you're not "another accounting tool," you're "the thing QuickBooks is bad at."

Failure mode: picking a competitor you don't actually know. The lists only work if you've used the product (or talked to 10 people who have).

Formula 4: The Regret

I have personally regretted not having [product] in the following 3 specific situations. So will 10,000 other people in similar situations.

Example: "I have personally regretted not having a way to send my contractor a $50 tip via text message at 11pm, in three specific situations in the last year. There are 30M US households who pay a home-services contractor annually and have the same problem."

Why it works: regret is a near-perfect leading indicator of demand. People don't regret things they didn't actually want. The "3 specific situations" requirement forces you to be honest about whether the regret is real or aspirational.

Failure mode: confusing regret with curiosity. "I wish there were a better tool for X" is curiosity, not regret. Regret has concrete memory attached to it. If you can't remember a specific moment, you don't regret the absence of the product.

Formula 5: The Copy + Twist

Take an existing successful product in one market. Copy its mechanics, twist the wedge for a different market.

Example: "Calendly is successful for sales calls. The same self-serve booking mechanic, applied to medical appointments, would let a patient book a 15-minute telehealth consult at 2am without calling the front desk. Same product, different buyer, different sales motion."

Why it works: the mechanics are proven; you're only taking risk on the new market, not on the product. The wedge is the specific buyer entry point — a particular specialty, a particular geography, a particular pricing tier.

Failure mode: underestimating how much of the original product's success was market-specific, not mechanic-specific. Calendly worked because sales reps had a high-volume, low-friction booking problem. Medical has a low-volume, high-friction problem — the mechanic has to change, not just the wedge.

Formula 6: The Unbundling

Take a product that does 5 things. Identify the one of those 5 things that the buyer uses 80% of the time. Build a focused product that does only that one thing, 10x better.

Example: "Salesforce does CRM, marketing automation, customer service, analytics, and app building. The median B2B SaaS company uses 80% of its Salesforce time in the CRM. A new product that focuses on the CRM wedge, with 10x the workflow speed and 10x the AI assistance, would be a credible $100M company."

Why it works: the buyer's behaviour has already told you which features matter. You're not guessing; you're observing. The "10x better" constraint forces you to be specific about what you're actually improving.

Failure mode: forgetting that the original product bundles those features for a reason (cross-sell, switching cost, support economies). An unbundled product has to solve for the standalone case convincingly.

Formula 7: The Boring Enterprise

Take a workflow that happens in every mid-size company in a specific industry, that is done with a spreadsheet and a Slack channel, that has not been modernised in 10+ years.

Example: "Every hospital system in the US has a 'nurse staffing coordinator' who, every shift, manually balances 200 nurses' preferences, certifications, and PTO against the next shift's patient census. This is done in Excel and a group text. The last software attempt was in 2011. A modern, mobile-first version of this tool, sold per-shift-scheduled, would replace the spreadsheet in 18 months."

Why it works: boring enterprises are the most reliable way to build a $10M–$50M ARR business with $0 in marketing. The buyer is captive, the pain is acute, the budget exists, and the incumbent software is usually 10+ years old. The "boring" part is your moat: most founders skip these because they're not cool.

Failure mode: underestimating the sales cycle. Enterprise sales is 6–18 months. You need either patience or a wedge (distribute free, charge for premium features) to survive the ramp.

Worked example: from vague to testable

Let's take a founder who says: "I want to build something for freelancers."

That's a category. Let's apply Formula 1.

"I help US freelance designers send invoices to enterprise clients that don't have a PO system in place, in less time than the 6 emails + 2 phone calls it currently takes."

Now apply Formula 2 (the painful workflow):

"Right now, a freelance designer has to email the client's AP department to get a vendor ID, then email again to confirm the PO format, then email a third time to ask why payment is 90 days late. I can reduce this to a single onboarding form that pre-populates the right AP team's contact details and PO template based on the client's industry and size."

Now apply Formula 3 (the two lists):

"FreshBooks does invoicing, expense tracking, and time tracking well. It does enterprise-AP integration, multi-currency reconciliation, and tax-form pre-fill badly. I'll build a focused tool that does only those three things, integrates with FreshBooks for the rest, and charges $40/month."

That paragraph is testable. You can run a 10-buyer-interview research sprint and find out within a week whether the workflow claim is real, whether the FreshBooks weaknesses are real, and whether the $40/month price is right. The original "freelancer tool" idea would have wasted three months before producing a single falsifiable claim.

The test for whether your idea is good enough

Before you spend a single minute on validation, run your idea through this checklist:

• You can name a specific buyer (a person with a specific job title, in a specific industry, at a specific company size).
• You can describe the job-to-be-done in one sentence without using the words "streamline," "leverage," or "optimize."
• You can quantify the current pain in time, money, or error rate, and you can cite a specific source for that number.
• You can describe a credible alternative the buyer uses today (a competitor product, a spreadsheet, a service, or "nothing").
• You can state a falsifiable claim: "I believe X buyers will pay Y dollars because Z."

If you can't tick all five boxes, the idea isn't ready for validation. Go back to the formulas. Pick a different one. Write the paragraph again.

What to do once the idea is testable

Once you have a paragraph that ticks all five boxes, your next step is not to build. Your next step is to test the falsifiable claim with the cheapest possible experiment. That almost always means 10 buyer interviews, not an MVP. Giri's 60-second validation report gives you the market-side research (competitors, trends, market size) but it can't tell you whether a specific buyer will pay. Only the buyer can tell you that. Run the interviews.

The seven formulas in this post are the upstream of that interview sprint. They exist to make sure the questions you're asking in those interviews are about an idea worth testing, not about a category worth hand-waving about.