Where Do B2B Startups Actually Find Their First Customers?

At seed, the channels that work are the ones where somebody already trusts you and the ones that reach people who are visibly living with the alternative you replace. Everything else is a test you are paying for twice, once in money and once in the quarter it takes to read the result.

Credibility you already hold is the cheapest distribution in the company, and it is not a channel in the media sense. It is the forty people who would answer an email from you by name: former colleagues, the customer who left for a bigger company, the introductions sitting unused in your inbox since March. Founders discount this because it does not scale. That is true, and beside the point. These first customers are the evidence you need before scaling is sensible to attempt, and they are the part of the job that does not transfer to anybody else yet, which I argued separately in what marketing a founder should do themselves.

What I would do with that list, concretely. Write it out as one page: forty named rows, one column for the person, one for what you think they use today instead of you. Work eight or ten rows a week, by hand. The first message asks one question and offers nothing to buy: what would you and your team do next month if the thing you use for this disappeared tonight.

Then the count. A reply is not a lead, and a compliment certainly is not. The raised hand is a named person who agrees to a second conversation with a colleague of theirs in the room, because that is the moment you learn there is a committee and who sits on it. Count those weekly. Let every other number wait.

The second source is where the replaced alternative earns its keep. Whatever your buyer would do next month if you shut down tonight is not only a positioning answer. It is a targeting instruction. The people doing that thing are findable, and where they are findable changes with what the thing is.

You replace a spreadsheet and a group chat

“The planner keeps the sheet. Everyone else asks her.”

Channel: the operator who owns the sheet, reached peer to peer. She can describe the pain in one sentence and has no budget. Your first job is to get her a number her boss can repeat.

You replace a system installed in 1994

“Nobody touches it. The person who configured it retired.”

Channel: the integrators and consultants already inside that estate, plus wherever IT and risk leads of that industry sit together. Your integration story travels; your interface does not.

You replace doing nothing

“We live with it. It has never been anyone’s project.”

Channel: outbound, named and specific, carrying the cost of another year of nothing. There is no community to join, because the problem has no owner yet.

3

One product, three first channels, and the only thing that changed was what the buyer would otherwise do. Pick the channel first and you are guessing which of these three you are in, with a media budget as the instrument.

Two things about that third column, the most common and the least served. A buyer replacing nothing has no deadline, no budget line and nobody accountable for the delay, which is why outbound into that situation has to arrive carrying the arithmetic of waiting rather than a demo offer. The group also tends to be larger than the first two combined, and it never appears in a competitor grid.

Start with organic, in the specific sense of conversations you can have by name this week, and keep paid for after a page has converted a stranger without you on the call. Paid is a volume instrument. It is excellent at buying reach for a sentence that works and indifferent to whether the sentence works.

Which is the thing to be careful about, and it is not a budget argument. Paid spend before the position is settled buys a larger audience for a claim nobody in the room can repeat: a bigger pile of people quietly filing you under the wrong category. You do not get a clean read back either. The campaign reports clicks and the committee reports silence.

What organic means at this stage is also narrower than the word suggests. It is not a content calendar. It is the founder having fifteen conversations, writing down what the buyer said in the buyer's words, and publishing the part that was useful to the buyer rather than flattering to the product. One written piece a week is plenty, and its raw material is that week's conversations, not a topic list. That work also produces the position, which is why I treat it as one activity and not two.

There is a benchmark worth reading sideways here. The 16th Annual KBCM and Sapphire Ventures SaaS Survey reports the mix of sales development work as 56 percent outbound, 21 percent inbound and 24 percent hybrid, averaged across 31 respondents whose median company holds $23.3 million of annual recurring revenue. Those companies have brands and a marketing function, and still do most of their sourcing by reaching out. My reading, not the report's: inbound is what a settled position produces later, not what a seed company lives on now.

The fastest way to get traction without burning your category is to concentrate every conversation on one segment that shares one replaced alternative, and to publish nothing that asks to be judged on a shelf you have not chosen yet. Speed is cheap. Speed in the wrong category costs a year: a category carries the evaluation criteria, the vocabulary and the price your buyer has in mind before you open your mouth.

Concentration is also what the early stage data points at, from a different direction. ICONIQ Growth's October 2024 study of 107 B2B SaaS companies scaling from $1 million to $20 million of annual recurring revenue reports that new logos account for 77 percent of gross new annual recurring revenue on average in the $1 million to $5 million band, on a cut of 115 datapoints, and that the companies crossing the plateau keep their focus on new logos. Almost all your growth at this stage comes from strangers. Strangers are precisely the people who cannot ask you what you meant.

This is where the category damage happens, quietly and at no visible cost. Each page you publish while the position is unsettled teaches somebody a shelf, and the pages do not agree with each other. In the five B2B companies between seed and Series A I reviewed recently, more than one live category was running at the same time in every single one. Nobody chose that. It accumulated.

So the sequence I would defend, before spending on reach, with the output named in each case. One written page that says what your buyers do instead of you, in their words. One list of named companies that share that alternative, built by hand; if you cannot get to thirty real names, the segment is not a segment yet. One category, held across every page you own, checked once a month by reading your home page, your deck and last week's outbound email in a single sitting.

Once the story converts, content carries the argument and paid buys reach for the pages that have already proved they work alone. The division of labour is strict: content is where the alternative is named and the value is stated in units the buyer already tracks, and paid is the amplifier pointed at the pages a stranger finished reading and acted on.

That order gives you a test instead of a hope. A page that converted twenty cold readers is worth spending on; a page nobody has read without a warm introduction is a hypothesis with a media budget attached. Run it the other way around and you end up with excellent cost per click and a pipeline that stops when the founder travels.

Which brings the real cost of waiting into view, and it is not the media spend. Channel spend is also category spend: every page you buy reach for teaches the market which shelf you belong on, and the shelf is the slowest thing in your company to change. A quarter of paid behind an unsettled story does not cost you a quarter. It costs you the year it takes to be re-filed.

An aside on who writes this. Building a go to market that holds before the spend is the work I do, for B2B founders selling to buying committees, from seed to Series A: twenty years installing it, four of them as CMO at HEMAV Technology for Xavier Silva, doing everything the project includes and more. The interview with the CEO I reported to is on the site. The Loom is a method that weaves a company's whole go to market into one story a buying committee can follow and buy: week one is a diagnosis drawn from your own calls, wins and words; by week four the positioning document is in your hands; weeks five to ten turn it into a go to market your team runs without me. It wins for the founder who wants the work to end, because one person sets the direction and does the building, it is priced as a fixed project rather than a monthly seat, and at week four you judge: if the document has not shown you the system, you cut the thread, the second payment never happens and you keep everything delivered.

Before you commit this quarter's spend to a channel, it is worth knowing whether the story that channel would carry holds up without you. The three things this piece asked of you are threads of a test I published. Knowing what you replace, in the buyer's words, is the one story thread. Getting the operator a number her boss can repeat is the value a finance lead can restate. One category held across every page is the materials thread, because materials only work if every function in the room reads the same company in them. The raised hands, worked into something that keeps arriving while you are on a plane, are the fourth: a pipeline that runs without the founder. A loose thread is a channel instruction in itself: if the value thread is the loose one, more reach only buys a wider audience for a claim your champion cannot yet carry into the room, so the spend waits until that thread holds.

Two of the twenty statements, exactly as the audit words them: “Our differentiation does not collapse into ‘faster, easier, cheaper’ when pressed, because we have a specific claim about where we genuinely win.” And: “I can give the board a pipeline number and trust that something close to it will actually land.”

Five statements on each of the four threads, rated from 1 (not true of us at all) to 5 (completely true), and the score builds in front of you out of 100. It lands in one of three bands and it names your loosest thread. Score each statement against your last real deal, not against what you meant to build. The audit is free, the rating takes a few minutes, and one email, given before you start, opens both the interactive audit and a printable PDF. Nothing else is promised.

Take the Committee Readiness Audit before you commit the budget, and spend against the thread it names.

Sources: 16th Annual KBCM and Sapphire Ventures SaaS Survey. ICONIQ Growth, Scaling SaaS from $1M to $20M ARR, October 2024. Xavier Silva interview, recorded July 2026, published at monkeyweaver.com/testimonial.