How to validate a
business idea

Idea validation is the discipline of testing whether a business or product idea is worth building, using real evidence. It includes defining the target customer and problem; interviewing customers about their current behaviour; listing assumptions and ranking them by risk; designing cheap experiments — landing pages, pre-orders, waitlists, concierge MVPs, prototypes; setting success criteria in advance; testing pricing; and deciding whether to build, change direction or stop.

The cost of skipping validation is high. CB Insights' March 2026 analysis of venture-backed shutdowns since 2023 found poor product-market fit cited in 43% of cases, and running out of capital in 70% — often after spending on products customers didn't want.¹ Even successful software companies build much that's never used: Pendo's 2019 Feature Adoption Report found 80% of features in a typical product are rarely or never used.² Product leaders at SVPG frame four big risks for any idea — value, usability, feasibility and business viability — which validation tests directly.³ Tools such as Strategyzer's Business Model Canvas help map the assumptions behind an idea.⁴

Startup shutdowns citing poor product-market fit (CB Insights, 2026)

43%

Shutdowns where capital ran out (CB Insights, 2026)

70%

Software features rarely or never used (Pendo, 2019)

80%

Big product risks: value, usability, feasibility, viability (SVPG)

4

Table: How idea validation skills are assessed (editorial summary)

Table: Mapping assumptions to cheap tests (illustrative)

Table: How idea validation practice developed (editorial summary)

• An interview script with non-leading questions
• Notes and patterns from 15+ interviews
• A ranked assumption list
• One experiment with pre-set criteria
• A results summary and decision
• A verified idea validation certificate

Show learning speed: "Validated a B2B idea with 25 interviews and a landing page; killed it in three weeks when criteria weren't met; pivoted to a related problem with five paying pilots." Honest stories of stopping are valuable.

"How would you validate this idea?"

Identify the target customer and riskiest assumption, design the cheapest test, set success criteria, run it and decide.

"What's wrong with asking 'Would you use this?'"

People are polite and poor at predicting behaviour; ask about past behaviour and ask for commitment.

"Your test results are mixed — what now?"

Check data quality and segments, refine the test or narrow the audience, and avoid moving goalposts.

"When would you stop an idea?"

When pre-set criteria fail repeatedly and no promising segment or variation appears.

"What's an MVP?"

The smallest thing you can build or deliver to learn whether your key assumption is true.

• "Who exactly is the customer?" — focus.
• "What must be true for this to work?" — assumptions.
• "What would convince me I'm wrong?" — criteria.
• "What's the cheapest way to find out?" — speed.

Table: Related roles and pay in India (AmbitionBox, October 2026; indicative)

Table: Common validation tools (editorial guidance)

• Founder — your own ventures.
• Product manager — discovery work.
• Innovation lead — corporate ventures.
• Growth manager — testing channels.
• Venture builder / investor — evaluating ideas.

1. Week 1: Interviews — Talk to ten target customers about the problem.
2. Week 2: Prove it — Run one cheap test with pre-set criteria; take the Idea Validation Certification.

Related: Entrepreneurship, Business Model Design and Pricing Strategy. Guides: entrepreneurship, business model design and product-market fit.

• Interviews — weekly customer calls.
• Assumptions — updating the list.
• Experiments — small tests.
• Data — reviewing results.
• Decisions — what to build next.
• Sharing — learnings with the team.

"What's a smoke test?"

Offering a product that doesn't exist yet — via a page or ad — to measure real interest.

"What's a pre-sale?"

Taking payment before the product is ready, often with a refund promise.

Table: A 30-day validation plan (editorial guidance)

Spend the first ten days talking to customers. Use the next ten to run a cheap test of the riskiest assumption. In the final ten, compare results with your criteria and decide.

AI can now generate landing pages, prototypes, survey analyses and interview summaries in hours. That makes testing cheaper and faster, but it also tempts founders to "validate" by asking AI what customers want. AI can help draft questions and spot patterns in notes; it can't replace real customers' behaviour. Use it to run more tests, not to skip them.

Open with context: "Tell me about your role and a typical day." Then ask about the problem area: "When did you last deal with X? What happened? What did you try?" Ask about costs: time, money, frustration. Ask what they use today and what they dislike about it. Only at the end, if appropriate, show your idea and ask for a commitment — a pilot, a pre-order or an introduction.

Before any test, write down what result would make you continue and what would make you stop — for example, "At least 5% of visitors pre-order at ₹220." Base targets on what the business needs to work, not on what feels achievable. Share them with a co-founder or mentor to keep yourself honest.

Pick a handful of target customers and deliver the service by hand — spreadsheets, calls, WhatsApp. Charge something, even a small amount. Watch where customers get value and where they struggle. Note which tasks repeat; those are what to automate first. Concierge tests teach more in two weeks than months of building.

"Friends liking the idea is validation." Only target customers' behaviour counts.

"You must build an MVP app." Many tests need no code.

"Validation guarantees success." It reduces risk; it doesn't remove it.

Tests often give unclear answers. Look at segments: did one type of customer respond far better? Check whether the test was fair — enough traffic, a clear offer, the right audience. If one segment shows strong signals, narrow your focus and test again rather than declaring the whole idea a success or failure.

• Leading questions.
• Testing easy assumptions first.
• No success criteria.
• Ignoring negative results.
• Building too early.

Startup failure data comes from CB Insights (March 2026); feature usage from Pendo's 2019 Feature Adoption Report; product risks from SVPG; and the canvas from Strategyzer. The lunch scene, assumption table and history are editorial. Science of Apps publishes the free test linked here. Last reviewed October 2026.

• Validation — evidence that an idea can work.
• Riskiest assumption — the belief most likely to kill the idea.
• MVP — minimum viable product.
• Concierge MVP — manual delivery to learn.
• Smoke test — measuring interest before building.