Customer Prospecting for B2B SaaS Founders

When you have only a few customers, your job is not to declare a definitive ICP. It is to identify the strongest repeated evidence and turn it into a prospecting hypothesis. A customer pattern can be useful even if it appears only twice, provided it is specific and tied to a meaningful outcome: the same job title, workflow, trigger event, implementation path, or reason for buying. Create a one-page record for every paying customer, recent trial conversion, and high-intent lost opportunity. Keep facts separate from assumptions. For example, “operations leaders at multi-location service businesses connected three existing tools within a week” is evidence. “All operations leaders at service businesses need this” is a hypothesis. This distinction prevents a small sample from becoming false certainty.

• For each account, capture: company size, industry, business model, buyer title, daily user title, champion title, trigger event, current workaround, alternatives considered, time to value, and the outcome they wanted.
• Mark each field as repeated, mixed, unknown, or one-off. Repeated fields are candidate targeting signals; one-off fields should not drive list building yet.
• Give extra weight to behavior over firmographics. A clear trigger or urgent workflow problem is usually more useful than a broad label such as “SaaS” or “mid-market.”

Not every similarity deserves a campaign. A useful prospecting pattern needs to be both plausible and findable. “Customers care about reducing manual work” may be true, but it is too broad to build a list. “Customers hire a revenue operations leader after moving from founder-led sales to a small sales team” is more actionable because you can identify companies and people with those attributes. Use a simple score before choosing a segment. Rate each candidate pattern from 1 to 3 for recurrence, pain intensity, ease of identifying prospects, and confidence that your product created value. Then subtract points for long sales cycles, difficult access to buyers, or a highly customized implementation. Start with the highest-scoring pattern, not the largest theoretical market.

• Example: three current customers are agencies with 20–80 staff, but only two share the same urgent need: client reporting is assembled manually after they add several enterprise accounts. If that trigger is visible through hiring, client announcements, or role changes, test “growing agencies with complex reporting” rather than “all agencies.”
• Decision rule: do not create a dedicated outreach list from a pattern that has no observable company signal and no reachable role. Keep researching it until you can name how you will find accounts and whom you will contact.
• Decision rule: if two customers look similar demographically but bought for different reasons, split them into separate hypotheses. Similar company profiles do not guarantee a similar message.

A focused outreach list is a written filter, not a collection of companies that vaguely look promising. Define the account criteria, the person criteria, the trigger, and the exclusion criteria before opening a database or browsing LinkedIn. This makes research faster and lets you later tell whether poor results came from the list or from the message. Keep the first list small enough to inspect manually. For a founder with limited evidence, 25 to 50 accounts is often enough for a meaningful first pass. You are trying to learn which signals predict a relevant conversation, not maximize sends. A narrow list also makes it possible to personalize around a real reason for contact without spending hours on every prospect.

• Use this template: “Companies in [specific category], with [size or operating characteristic], showing [trigger], where [role] likely owns [workflow], excluding [disqualifier].”
• Hypothetical example: “US-based vertical SaaS companies with 30–150 employees that recently hired their first customer success leader, where the CS leader owns onboarding consistency, excluding companies that already advertise a dedicated customer onboarding platform.”
• Set boundaries in advance: geography, employee range, funding or maturity stage if relevant, technology signals only if they affect the workflow, and roles to exclude. Exclusions are as important as inclusions because they stop obvious mismatches from consuming outreach capacity.

Build accounts first, then map people. Starting with names often creates a scattered list of job titles across companies that have no shared reason to care. For every account, record the evidence that qualified it: a hiring post, product launch, new location, stated workflow, job description, customer type, or public technology signal. If you cannot explain why an account belongs in one sentence, it should remain unqualified. Next, identify the likely buying group. In small B2B sales, the economic buyer, workflow owner, and internal champion may be different people. Contacting only a senior executive can hide useful feedback; contacting only an operator can stall a deal. Choose one primary role based on who feels the pain most directly, then add one secondary role when the account is important enough to justify it.

• For each account, save: qualification evidence, primary contact, secondary contact, likely workflow, possible trigger date, current tool or workaround if known, and a disqualifying signal if discovered.
• Use a three-level status: qualified means the account matches your definition and has evidence; watch means it is plausible but missing a trigger or contact; reject means it violates an exclusion or has an incompatible workflow.
• Puffle helps teams find relevant people, research them, and prepare email and LinkedIn outreach for review. Use that workflow to keep the reason for inclusion attached to each person instead of treating contact records as detached names.

Your first outreach should test a specific connection between the observed trigger and the problem you solve. Do not pretend to know an internal problem from a public signal. State what you observed, describe the plausible operational consequence as a question or conditional, and offer a short reason to compare notes. This is more credible than claiming that every company with a certain title or funding event has the same need. Make one message variable at a time. If you change the segment, trigger, value proposition, call to action, and channel simultaneously, replies will not tell you what worked. Run one tightly defined batch, review the quality of responses, then adjust. A negative reply can still validate your targeting if it reveals the wrong owner, timing, or competing priority.

• Hypothetical email structure: “I noticed [company evidence]. Teams at this stage sometimes run into [specific workflow consequence] when [trigger]. We help [role] handle [job] without [current workaround]. Is [workflow] something you own, or is someone else closer to it?”
• For LinkedIn, use the same premise but reduce it to the observed event and one question. Do not send a generic connection request followed immediately by a pitch.
• Decision rule: use a direct meeting request only when evidence is strong. When evidence is weaker, ask a routing or diagnostic question first. The goal is a relevant response, not a forced calendar link.

Assess a list using conversation quality, not just volume. A campaign that produces few replies but several clear confirmations of a painful workflow may be worth refining. A campaign that produces many polite replies from companies that will never buy is not a success. Review account fit, contact accuracy, relevance of the trigger, message clarity, and objections separately so you do not blame copy for a weak segment. Keep a learning log after each batch. Record what prospects said in their own words, which titles engaged, which assumptions were wrong, and why accounts were rejected. After 25 to 50 well-researched accounts, decide whether to expand, revise, or stop the hypothesis. Expansion should mean finding more accounts with the same evidence, not loosening every filter to fill a larger list.

• Expand when at least several relevant prospects confirm the problem, the intended role is reachable, and the implementation or buying motion resembles successful customers.
• Revise when prospects recognize the problem but assign it to another role, describe a different trigger, or use language that makes your message feel off-target.
• Stop when qualified prospects consistently lack urgency, have an incompatible workflow, or require product capabilities that your current customers did not need. Stopping early protects founder time and produces a cleaner next hypothesis.

If you want help doing this work, these Puffle pages show the product in more detail.

• How Puffle works
• Outbound automation
• Browse all GTM guides