Compare methods by the evidence they create
| Method | Best question | Evidence strength | Main limitation |
|---|---|---|---|
| Desk and competitor research | Does a market and buying category exist? | Context | Existing supply does not prove demand for your offer |
| Problem interviews | How does the buyer experience and solve the job now? | Qualitative | Stated pain is not purchase behavior |
| Survey | How common is a known pattern in a defined sample? | Directional | Sampling and hypothetical-answer bias |
| Search and community research | Is there observable language and attention around the problem? | Directional | Attention may not convert at your price |
| Landing-page or fake-door test | Will the audience act on this proposition? | Behavioral | Depends heavily on traffic quality and disclosure |
| Waitlist | Will a visitor exchange contact details for follow-up? | Low-cost intent | Email capture has little commitment |
| Pricing interaction | Does attention survive seeing the price? | Commercial intent | An interaction is not a purchase |
| Reservation or deposit | Will the buyer accept meaningful friction? | High intent | Creates trust, operations, and possible legal obligations |
| Concierge pilot | Will the buyer pay for the outcome before automation? | Commercial delivery | Manual service may hide software economics |
| Working MVP | Can users activate and return to a product? | Product behavior | Highest cost and still not proof of a scalable market |
Use methods in order of irreversible cost
Validation should reduce uncertainty before the next expensive commitment. The sequence below is a default, not a ritual; skip a stage only when you already have equivalent evidence.
Map the market
Identify the buyer, purchasing process, current alternatives, category language, and reachable channels. This stops an interview sample from defining the whole market by accident.
Understand recent behavior
Interview people about the last time the problem occurred, what it cost, what they tried, and who approved spend. Past behavior is more useful than forecasts about a hypothetical app.
Test one proposition
Put a specific promise and price in front of qualified traffic. Instrument exposure and progression through the offer rather than treating every visit as equivalent.
Increase commitment
Move from a low-cost CTA to a detailed application, reservation, deposit, or paid pilot only when a stronger signal is necessary and you can honor the promise.
Deliver the outcome manually
For high-value workflow software, a concierge pilot can reveal whether the outcome is valuable before automation. Track labor and exceptions so service demand is not mistaken for viable software margins.
Build the smallest activation loop
Only after acquisition and commitment evidence, implement the shortest product loop that produces a real customer result. Retention is a separate test from pre-launch demand.
Choose the method from the risk, not the trend
| Uncertainty | Start with | Escalate to |
|---|---|---|
| Buyer has no urgent problem | Recent-behavior interviews | Concierge problem-solving session |
| Audience may be unreachable | Channel and list research | Small controlled acquisition test |
| Promise may be unclear | Message interviews | Landing-page proposition test |
| Price may be rejected | Explicit price presentation | Reservation, proposal, or paid pilot |
| Buyer may not trust delivery | Process proof and manual demo | Narrow design-partner engagement |
| Solution may be technically infeasible | Technical spike | Constrained prototype with real inputs |
| Users may not return | Concierge repeat workflow | Instrumented activation and retention MVP |
If the main uncertainty is which market deserves a test at all, score candidates using pain, purchasing power, and reach. The “what SaaS should I build?” framework gives you a structured shortlist instead of a generic list of app ideas.
The sample can invalidate an otherwise good method
Ten relevant buyers can teach you more about a workflow than a thousand convenience respondents. This does not make small samples statistically representative; it means relevance and traceability matter before scale.
- Define qualification before recruiting: role, company type, workflow, geography, and recent problem exposure.
- Record how each participant or visitor was recruited and which message brought them in.
- Separate the user, internal champion, budget owner, and procurement gate when they differ.
- Do not combine paid social, founder followers, and direct outreach without analyzing the source cohorts.
- Avoid incentives that are more valuable than the action you are trying to interpret.
- Report uncertainty and missing segments rather than decorating a small sample with precision.
Know what pre-launch validation cannot prove
That boundary is useful. The purpose of validating a SaaS idea is to avoid funding a product before a reachable market accepts the proposition. Once that evidence exists, move to delivery, activation, retention, and unit economics with the same discipline.
For an end-to-end behavioral protocol, read how to validate a SaaS idea before building. For stronger commercial signals and their obligations, read how to pre-sell SaaS transparently.