How do you generate leads when you don’t even fully know your ICP yet?

1. Start with a working ICP, not a perfect one

You don’t need to magically figure everything out before you start. Even when your ICP feels blurry, you already have some signal. The problem is pretending that signal needs to be perfect before you act.

What you need is a working ICP, a strong hypothesis. Something good enough to move, not something polished enough to present in a deck. 

• Keep it intentionally rough. 
• Pick a broad industry, not a hyper-specific niche. 
• Define a realistic company size range, not your dream customer. 
• Anchor everything around one problem you know you can solve right now.
• Then choose the role that feels closest to that pain, even if you’re not sure they’re the final decision-maker.

No personas yet, no positioning exercise, and definitely no overthinking. This is about having something concrete to put in front of real people. Once you do that, behavior does the work for you. Who engages, who asks questions, who follows up, who disappears. That’s the data that sharpens your ICP. 

2. Run “learning-focused” lead gen, not growth campaigns

When your ICP isn’t clear, the goal of lead generation changes. You’re not trying to scale yet, you’re trying to learn. That means you optimize for conversations and patterns, not MQL volume or cost per lead.

At this stage, good channels are the ones that give you fast, qualitative feedback. You want to see who replies, what they push back on, and where interest drops.

For B2B, good channels at this stage are the ones that give you direct, qualitative feedback:

• LinkedIn outbound with one or two tightly framed messages tied to a specific problem
• Founder-led or expert-led cold email, where the message reflects real experience instead of generic copy
• High-intent content with a narrow CTA, like an audit, teardown, diagnosis, or focused consult
• Warm intros and partnerships that shortcut trust and surface real objections quickly

For B2C, good channels look different. You’re still optimizing for learning, but learning comes from behavior more than conversation:

• Small-budget paid social tests with tightly scoped audiences and one clear message
• High-friction offers that require intent, like short assessments, waitlists, or paid trials
• Direct response landing pages with one action and clear problem framing
• Email capture tied to a specific outcome, not generic discounts or newsletters

In both cases, the common thread is fast feedback and clear signals.

Bad channels also differ, but the pattern is the same: anything that prioritizes volume before understanding.

For B2B, bad channels at this stage include:

• Broad paid ads that force you to spend before you know what actually converts
• SEO at scale, where feedback cycles are too slow to help shape an early ICP
• Generic lead magnets designed to capture as many emails as possible

For B2C, bad channels include:

• Wide-reach paid campaigns optimized for clicks instead of post-conversion behavior
• Influencer campaigns before you know which message or audience resonates
• Discount-led offers that attract price-sensitive users without teaching you anything

If a channel makes it harder to understand why someone converted or didn’t, it’s the wrong channel right now. You can scale later. First, you need signal.

3. Offer something that filters people

At this stage, what you offer is doing most of the work. And it shouldn’t be designed to appeal to everyone. Your offer should filter people out as much as it pulls the right ones in.

If your lead magnet sounds useful to anyone with a website or a business, it’s too generic. You’re not learning anything from that. You want an offer that makes some people think, “this is exactly for me,” and others immediately bounce. That’s a good sign.

Be very clear about the context and the limits. Say who it’s for and who it’s not. For example:

• “Free 15-minute teardown for B2B SaaS websites with more than 10k monthly traffic.”
• “We only work with X. If that’s not you, this won’t be useful.”

That kind of framing feels uncomfortable at first, but it does something important. It raises intent for the right people and cuts noise fast. You’re not trying to maximize sign-ups. You’re trying to see patterns.

If everyone converts, your ICP is too broad and your offer isn’t filtering enough.  If no one converts, your problem framing is probably off, or you’re aiming at the wrong pain.

Both outcomes give you signal. And at this stage, signal is more valuable than volume.

4. Track who converts and why, not just that they convert

At this stage, it’s not enough to know that someone converted. You need to understand who converted and why they bothered.

A booked call or a form fill is just a signal that something clicked. On its own, it’s pretty useless. What matters is what pushed that person to act in that moment, what problem they had in their head, and what nearly stopped them.

Early on, your CRM shouldn’t feel like a reporting tool. It should feel like a notebook. You want to capture who booked the call, what role they have, what triggered them to reach out, how they described the problem in their own words, and where they hesitated or pushed back.

After a handful of real conversations, patterns start showing up. The same roles. The same frustrations. The same questions before pricing comes up. You’ll learn more from ten honest calls than from a thousand impressions ever could. That’s the kind of data that actually shapes your ICP.

5. Narrow aggressively after each batch

This is usually the moment where teams get uncomfortable, because it forces you to let go of ideas you were attached to.

After every small batch of conversations, roughly ten to fifteen, you should go back and reassess what you’ve just learned. Not in a formal way, and not with a complex framework. Just by looking at the conversations and being honest about what actually happened.

Some people show urgency immediately. Others are interested but clearly not in a position to act. Some grasp the value within a few minutes, while others need long explanations and still don’t quite get there. Some ask about pricing early because they’re already evaluating options. Others stay in education mode and never really move forward.

When you look at those patterns together, your ICP starts narrowing on its own. Certain roles stop appearing. Certain use cases drag every conversation out. One or two types of companies consistently move faster and feel easier to sell to.

6. Only then do you scale lead generation

Scaling lead generation only makes sense once learning has done its job. Before that point, scaling doesn’t create more opportunities, but more noise, more spend, and more confusion.

You know you’re ready to scale when lead generation stops feeling random. When conversations start sounding familiar and objections repeat. When certain types of prospects move faster, ask better questions, and need less convincing. That’s when lead gen shifts from exploration to execution.

At that point, it’s about amplifying what you already know works.

Paid channels start working because you’re no longer guessing what to lead with. You know which problem triggers action, which role reacts to it, and what language resonates. Instead of testing five vague messages, you can put budget behind one or two proven angles and measure performance with confidence. Paid spend stops being an experiment and starts behaving like a lever.

SEO also changes role. Earlier on, content is useful for learning and credibility, but slow for feedback. Once your ICP is clear, SEO becomes a compounding engine. You’re no longer writing broadly about a topic. You’re creating pages around specific pains, comparisons, and use cases you already know convert. That’s when organic traffic becomes qualified, not just abundant.

Messaging improves across the board. Your homepage becomes clearer because it’s written for a specific buyer. Lead magnets convert better because they speak to a real situation, not a generic need. Outbound becomes more efficient because you’re repeating language that already worked in conversations, not inventing new copy every week.

Scaling also means tightening the funnel, not just widening the top. You start filtering harder. You raise friction intentionally. You say no to segments that slow deals down, even if they technically could buy. This is where lead generation becomes selective instead of inclusive.

Scaling before clarity doesn’t accelerate growth. It delays it by making learning more expensive and harder to interpret.