How to Validate a SaaS Idea Before You Build It

To validate a SaaS idea, collect evidence that a specific group of buyers has a painful problem and will give up something real to solve it, before you write production code. NetSuite describes it as replacing guesswork about demand with evidence from customer conversations, sign-ups, preorders and, ideally, real money changing hands. This guide gives you the full routine in seven steps. It also shows where a structured outside read on your concept fits in once you have something to show. At VibeVoted, ten leading AI models score software entries on seven published criteria, and those criteria map closely onto the questions validation has to answer.

TL;DR

• Validation is evidence of behavior, not encouragement. Compliments and email addresses are weak. Deposits, signed agreements and booked calls are strong.
• Work in order: pick a narrow buyer, prove the problem, study paying competitors, test the promise, then ask for a commitment.
• Cheap demand tests (landing page, fake door, concierge, pre-sale) can run for well under the cost of a build.
• Decide in advance what counts as pass, iterate or stop. Write the kill criteria before you see the data.
• Once you have a page, prototype or live product, a VibeVoted voting session returns scored, model-written feedback on seven criteria for 25 USD per credit.

Table of Contents

• What validating a SaaS idea actually means
• The seven steps, in order
• What to measure once an MVP exists
• Add a structured outside read to your evidence
• Put your page in front of ten AI models
• A worked example
• Common mistakes to avoid
• FAQ
• Sources

What validating a SaaS idea actually means

Most people use the word to mean encouragement. They describe the idea to ten people, eight say it sounds useful, and they call it validated. PlanMySaaS points out the flaw: those eight were being polite, were asked to give up nothing, and had not tried to solve the problem themselves. Its definition of validation is a sequence of cheap tests that try to kill the idea in the order that kills it fastest. A test that cannot produce a “no” is not a test.

SaaS raises the bar for three reasons, according to 1mil.app:

1. Subscription revenue. You need evidence of repeat use, not a single enthusiastic yes.
2. Slow, costly builds. Evidence gathered before coding matters more than it would for a service business.
3. Feature risk. Your idea may be a feature of someone else’s product, not a standalone budget line.

If your product runs on AI models, add a fourth. NetSuite notes that AI products tend to carry thinner margins because every model call costs money, so validation now includes modeling the economics, not only the demand.

The seven steps, in order

Step 1: Define the decision and the buyer

Start by writing down what you are trying to decide. The usual answer is “build, change direction, or stop.” TestBeforeYouBuild frames its whole process around that three-way choice and argues that an MVP is a tool for testing particular assumptions, not the first move for every idea.

Then choose one narrow segment. “Small businesses” is not a segment. “Three-to-ten-person recruiting agencies that track candidates in spreadsheets” is. In B2B, name the roles too: the person who feels the pain, the person who holds the budget, and anyone in security or procurement who can block a purchase.

Where it goes wrong: a broad audience makes every conversation sound encouraging and teaches you nothing. If different segments describe different core problems, you have not found your buyer yet.

Step 2: Write assumptions you can falsify

Turn the idea into statements that data can prove wrong. For example: “At least half of the agencies I speak to will describe a workaround they built themselves” or “Three of ten qualified prospects will pay a deposit.” Set the thresholds now, before you see results, so you cannot move the goalposts later.

ZeroRaised splits the work into four kinds of evidence, which makes a handy checklist:

Step 3: Interview for the problem first

Talk to people before you show them anything. Reforge recommends 10 to 30 conversations to build conviction, split into two types: long-form interviews that explore the problem area without priming the person, and short take-away conversations of about 10 minutes that test interest in your proposed solution. Keep your questions open-ended so people describe the problem in their own words.

Ask about the last time the problem happened. What did they do? What did it cost them in time or money? What have they already tried? Past behavior is evidence. Opinions about a hypothetical product are not.

Track the replies. Reforge suggests counting the yes and no answers to direct interest questions, which keeps you honest when a friendly conversation tempts you to remember only the good parts.

Where it goes wrong: pitching in the first five minutes. Once the person knows your idea, they start being kind. TestBeforeYouBuild also warns that positive feedback, sign-ups and feature requests are encouraging but not decisive on their own. Stronger evidence connects to behavior: repeated problems, current spending, a reason to change and a realistic way to buy.

Step 4: Study the market and the paying competitors

Competition is usually good news. Valuflash argues that a market with paying competitors is often evidence of real demand, and that the sharper question is what they do poorly and for whom. Stripe treats competitive analysis as a part of market research: who else is solving your customers’ problem and where they fall short.

A practical routine:

• List every product your buyer uses today, including spreadsheets and manual processes.
• Read their public pricing to see what buyers already pay.
• Read recurring complaints in reviews and forums. Complaints show where the market is underserved, and repeated praise shows what you cannot casually leave out.
• Look at who bids on the keywords tied to your problem. Heavy ad spend from established players suggests customer acquisition will be expensive.
• Check search trends over time, not a single spike. A burst of social chatter is not the same as a baseline of demand.

Where it goes wrong: copying a competitor’s feature list. Having a feature in a rival product does not prove buyers want it, and obsessing over rivals can leave you a follower with no differentiation. Equally, an empty field usually means no market or a customer cost too high to sustain.

Step 5: Test the promise cheaply

Now show something. The cheapest honest tests are:

• Landing page smoke test: a page with a clear promise and one call to action.
• Fake door test: a button for a feature that does not exist yet, to count clicks.
• Concierge MVP: you deliver the result by hand for a few customers.
• Wizard-of-Oz prototype: the interface looks automated while you work behind it.
• Pre-sale: you ask for payment or a signed letter of intent.

SpeedMVPs puts each of these at roughly $0 to $500 and anywhere from hours to two weeks. Startupik says a basic smoke test can cost under $100 with organic outreach, a no-code page and free analytics, while a paid-traffic test usually runs $200 to $1,000. It adds that B2B and developer tools tend to need more because qualified clicks cost more.

Each test answers only part of the question. ZeroRaised notes that a landing page shows a promise attracts attention but not that people will pay, a pre-sale shows a few will pay but not that you can acquire customers affordably, and interviews show how the problem feels but not how big the market is. For a solo maker its suggested order is interviews first, community research alongside, then a landing page, then a pre-sale to the people who responded.

Step 6: Ask for a commitment

This is the step most founders skip, and it is the one that matters. A waitlist sign-up is weak evidence. A booked call is real interest. A prepayment or a paid concierge engagement is the closest thing to proof. NetSuite makes the same point: a signed agreement or deposit beats a compliment or an email address.

Scale the ask to the stage. For a consumer-style tool, that might be a paid early-access price. For B2B, it might be a letter of intent, a paid pilot, or a meeting with the person who controls the budget. If you cannot get ten people to take a real action without begging, the problem is probably the product, not the funnel.

Step 7: Compare evidence against your thresholds and decide

Go back to the numbers you wrote in step 2 and make one of three calls: build, change or kill. ProofEngine offers a workable frame. Treat “pass” as proceed, “conditional” as investigate further and “fail” as the assumption did not hold. Its stop signals include most experiments missing their thresholds, nobody clearly articulating the problem, and zero demand across channels. Its signals to iterate include pockets of interest, customers describing a different problem than you expected, or one segment responding far better than the rest.

Two red flags deserve special attention. The first is people saying yes but not acting. The second is the “vitamin” response, where prospects agree the problem exists but say they are managing fine. Bright Curios adds that if fewer than four in ten qualified prospects describe an active workaround, you should review your direction.

A stop is a good result when it comes early. Learning in a few weeks that a market is not there is the process working, compared with learning it after months of development.

What to measure once an MVP exists

Evidence does not end at launch. The MVP is itself a test of specific hypotheses, and the metrics you track should say whether those hypotheses held. VL Studio groups them into four families: activation (core action completion and time to first value), engagement, retention (cohorts and churn) and business metrics (conversion, CAC, LTV and the LTV:CAC ratio).

MVP Development Company argues that activation and retention matter most early on. Activation tells you the product works. Retention tells you people want it. Acquisition and referrals come second, because pouring users into a product that does not retain only fills a leaky bucket faster. Because SaaS earns through subscriptions, retention is the number that most directly tests the idea you started with.

Add a structured outside read to your evidence

Customer conversations tell you about the problem. Once you have a landing page, a prototype or a live product, you also need to know how a stranger reads it. Does the promise come across? Does the page look trustworthy? Does it differ from what exists? Founders are too close to their own work to judge this reliably.

That is where our approach helps. VibeVoted ranks software products using evaluations from 10 distinct leading AI models. Each entry is scored on seven published criteria: clarity, need, differentiation, evidence, visual design and polish, trust, and sustainability. Read them next to the four evidence types above and the overlap is plain. Need and evidence speak to problem and solution validation. Differentiation speaks to the competitor gaps from step 4. Trust and sustainability speak to whether a buyer would commit and keep paying. Clarity and visual polish speak to whether your landing page test is fair, since a confusing page can sink a good idea.

Every entry receives model-generated feedback, and the leaderboards show your product side by side with others in its category. Our rankings rest on repeatable AI model evaluations across fixed criteria, not on user voting or popularity campaigns, and sponsor payments buy exposure without influencing placement. You can read the rules and methodology before you decide whether it fits your stage.

Use it as one more input alongside interviews and pre-sales, ideally when your page is live and before you spend on paid traffic, so that feedback on clarity and differentiation can shape the copy you test.

Put your page in front of ten AI models

A voting session costs 25 USD one-off per credit. Software makers enter their product, get it ranked against others in the category, and receive written feedback from each model. If you have a landing page or a working prototype, you can sponsor a voting session and see how the models score clarity, need, differentiation and trust. Browse the current leaderboards first to see how products in your category are scored.

The weekly, monthly and all-time rankings show what high-scoring entries look like, which is useful reference material for step 4 as well.

A worked example

Say you want to build a tool that turns support tickets into a weekly product insights digest for small SaaS teams.

1. Segment: product managers at SaaS companies with five to twenty employees who read support tickets by hand.
2. Assumptions: at least half describe a manual workaround; at least three of ten will pay a deposit.
3. Interviews: you speak to fifteen people. Nine describe a spreadsheet or tagging routine, and two say they are managing fine.
4. Competitors: you find broad analytics suites that are priced for larger teams and many complaints about setup time. That is a gap for small teams, and paying competitors show the market spends money here.
5. Test: you run a landing page and a concierge version, writing the first three digests by hand.
6. Commitment: four of the fifteen agree to a paid pilot at a low monthly price.
7. Decision: the pass threshold was three, so you build a narrow version and measure whether pilot users still open the digest in week four.

Nothing here required code, and every step produced a number you can compare with the threshold you set in advance.

Common mistakes to avoid

• Treating politeness as proof. Friends and colleagues say yes to be kind.
• Building first and researching later. An MVP tests assumptions. It does not replace having them.
• Skipping the economics. A product that earns per subscription but costs per model call needs a margin check early.
• Confusing interest with intent. Clicks and followers are signals of attention, not budget.
• Moving the goalposts. Without thresholds written beforehand, any result looks like a pass.

FAQ

How long does it take to validate a SaaS idea?

A focused round of interviews, competitor research and one cheap demand test can fit into several weeks. The cheap experiments in the research run from hours to about two weeks each. Plan on a few weeks of evidence gathering, and stop earlier if your kill criteria trigger.

How many customer conversations do I need?

Reforge suggests 10 to 30 conversations to build conviction, mixing long interviews with short follow-ups. What matters more than the count is that the people are qualified, belong to one segment and describe real past behavior. Keep a tally of yes and no answers.

What is the best sign that people will pay?

Behavior that costs them something: a deposit, a pre-order, a signed letter of intent or a paid pilot. Waitlist sign-ups and compliments rank far below that. Ask for the commitment as early as you can show a credible promise.

Should I build an MVP to validate the idea?

Not necessarily first. Before production code, you can study the customer, the workflow, existing alternatives and the buying process. Build an MVP when you need to test a particular product assumption that research cannot answer, and decide in advance which metrics will tell you it worked.

How can VibeVoted help me validate a SaaS product?

Once you have a page or product, a voting session returns scores from 10 AI models on seven criteria, plus written feedback, for 25 USD per credit. It adds a structured read on clarity, need, differentiation, evidence, polish, trust and sustainability to the evidence from your customers.

Sources

• NetSuite: How and Why You Should Validate Your SaaS Startup Idea
• PlanMySaaS: How to Validate a SaaS Idea
• 1mil.app: How to Validate a SaaS Idea Before Writing Code
• TestBeforeYouBuild: How to Validate a SaaS Idea Before Building an MVP
• ZeroRaised: How to Validate a SaaS Idea Before Building
• Reforge: Validate big bets with customer interviews
• Valuflash: SaaS Idea Validation
• Stripe: How To Do Market Research For a Startup
• SpeedMVPs: How to Test Your MVP Idea Before You Build
• Startupik: How to Run a Smoke Test for Your Startup Idea
• ProofEngine: Validation Kill Criteria
• Bright Curios: Pivot Signals for Bootstrapped Founders
• VL Studio: How to Measure MVP Success
• MVP Development Company: MVP Metrics