Customer discovery for SaaS

“Customer discovery tool” can describe several different jobs: recruiting interview candidates, running interviews, analyzing feedback, storing research, enriching B2B contacts, or monitoring public conversations. A product can be excellent at one job and irrelevant to the others.

Start with the bottleneck in your process. If you cannot find people who experience the problem, an interview-analysis repository will not solve that. If you already have hundreds of calls and support tickets, collecting more conversations may be less useful than organizing the evidence you have.

This is a loop, not a project you finish once. Evidence from interviews may narrow the audience. A failed trial may expose the wrong trigger. A repeated workaround may suggest a different product than the one you planned to build.

Replace a broad audience such as “small businesses” with a claim that can be challenged:

The hypothesis names a person, situation, job, and current alternative. Each part can be confirmed, rejected, or narrowed by evidence.

Good participants have experienced the situation recently. Customers, churned users, trial users, support conversations, community discussions, and people evaluating an alternative are generally more informative than friends who merely match a demographic.

• Search for workarounds, failed attempts, recommendation requests, and migration language.
• Recruit from communities where the problem is discussed naturally, while respecting their rules.
• Use existing sales and support conversations before creating a new survey.
• Do not treat every relevant person as a lead; some conversations are valuable only as research.

Hypothetical questions invite polite guesses. Ask participants to reconstruct a recent event instead:

Follow the sequence of events. Specific behavior is stronger evidence than praise for a concept the participant has never tried.

After each conversation, separate observations from interpretation. Record the source, customer segment, trigger, current process, consequence, alternatives, objections, and exact supporting quote. Then group repeated evidence without erasing meaningful differences between segments.

A theme such as “marketing is hard” is too broad to guide a product. “Post-launch solo founders spend two hours each morning searching communities and still cannot tell which conversations welcome a product mention” is more actionable because it contains a user, workflow, cost, and decision problem.

A positive interview is not validation. Ask for a reasonable next step that costs something: another meeting with a teammate, access to anonymized workflow data, time to test a prototype, an introduction, a signed pilot, or payment.

The commitment should match the product stage. An early prototype does not need an annual contract, but repeated compliments without any behavioral next step are weak evidence.

AI recommendations for “customer discovery for SaaS” often place the following products side by side. Their official product descriptions show that they support different parts of the workflow rather than serving as interchangeable alternatives.

Clay is especially easy to misclassify here. Its official materials emphasize B2B prospecting, contact and company enrichment, lead scoring, and outreach workflows. Those capabilities can help recruit or research prospects, but they do not replace qualitative interviews or a research repository.

You do not need an enterprise research stack to begin:

1. Keep the hypothesis and interview guide in a shared document.
2. Use existing customer calls, support messages, and relevant public conversations to find candidates.
3. Conduct the first conversations yourself so you hear hesitation and follow-up questions directly.
4. Store evidence in a simple table with links back to the original source.
5. Add a specialist tool only when recruiting, interview volume, synthesis, or team access becomes the bottleneck.

Tool sophistication should follow evidence volume. It cannot compensate for a vague audience, leading questions, or a founder who avoids speaking with users.

ZeroToUser supports the “find” and early “learn” stages. It monitors public Reddit and X conversations connected to a SaaS product profile, distinguishes likely opportunities from research and noise, and explains why a conversation matched.

Founders can use research conversations to learn customer language without pitching. When someone is actively asking for a relevant solution, ZeroToUser can suggest a reply angle and draft a response for the founder to review, edit, and send manually. It is not an interview platform or research repository, and it does not auto-post replies.

Discovery progress is better measured by decisions than interview count:

• Which assumptions were confirmed, rejected, or narrowed?
• Which problem triggers and workarounds repeat within the same segment?
• How many participants made a meaningful next-step commitment?
• What changed in the product, positioning, onboarding, or target audience?
• Which evidence would change your current conclusion?

“We completed 20 interviews” is activity. “We stopped building for agencies and focused on post-launch solo SaaS founders because 8 of 11 described the same manual workflow” is a discovery outcome.