How do I build demand before my startup has a well-known brand?

An unknown name can still earn a next step

You build demand before a brand by getting one specific buyer to take a costly next step. Logo recognition is a later shortcut for trust, not a prerequisite.

Paul Graham wrote in 2013 that founders make a startup take off by recruiting users by hand. You cannot wait, and you cannot hire the early sales away. A known brand already answers who you are and whether this will work. An unknown company answers that in the conversation, on the page, or in work the buyer can inspect.

Startupik separates attention from demand. Demand means someone will risk time, reputation, money, or a change in how the work gets done. A like, a page view, or “tell me when you launch” is not that risk.

Brand, category demand, and captured intent are three different jobs

Name recognition, category demand, and captured intent are three jobs, and only captured intent assumes people are already looking. Commitment is the step you count, not a fourth awareness job.

Creating demand before capture says lead generation only harvests people already searching, and category content should still help if the product did not exist. Demand generation for startups splits the rest into attention, then a meeting a person owns. The failure is the gap.

Job	Buyer already has	Move	Wrong first spend
Name recognition	No belief about you	Person, peer, or a checkable result	Identity project or launch week
Category demand	Pain, no name for a fix	Name the pain and the workaround’s cost	Demo ads, gated sheets
Captured intent	A search or a shortlist	Answer it, offer one next step	Broad awareness
The step you count	Enough trust to act	Call, file, pilot, or introduction	Followers or a raw list

That strategy guide later waits for fit signals before a demand program. Read it as a limit on always-on spend. Conversations start earlier, because they teach the frame. Paid awareness waits until a buyer repeats the sentence. Do not hire the ad retainer to invent the wedge.

Pick one motion from the buyer’s situation

Pick the motion from how this buyer can be reached, and from the commitment you can count. Refuse the others until that one produces a step or a clean no.

Motion	Use when	Leave when	Count this
Founder conversations	One meeting can move a named buyer	A call will not pay back	They restate the problem and book
Problem content	Buyers research in public	Only other founders reply	A stranger asks or forwards
An existing room	The buyer already gathers with peers	No practitioner room, or you are the only member	They ask you into their workflow
Warm introduction	A non-competitor already speaks with this buyer	The intro would borrow a logo you have not earned	They make the introduction
Manual delivery	You can do the outcome by hand once	Unsafe by hand, or value only at volume	They share material and ask again
Segmented waitlist	You will qualify each join	The email count is the goal	A segment names a use case
Search or retargeting	People already search or visited	You are still teaching the frame	A query tied to a conversation

Graham says a big launch and a large partnership usually fail to start growth, because both skip one-by-one recruiting. Facebook began at Harvard: density beats a thin presence everywhere. Graham says the best early adopters among companies are usually other startups, because they are more open to new things and have not locked their choices yet.

Adwave says outreach often beats paid for first B2B customers, and search only catches demand that exists. It also sells television to look established, against its own direct-response rule. Founder-led growth leads with sales and content in B2B, and with community plus product behavior in consumer sales. Outbound uses 30 to 50 named accounts and 15 to 20 personal notes a week. After 60 to 80 empty messages, change the list or the sentence, not the tool. See founder-led GTM before a full team and outbound for an early-seed B2B startup.

In the Netherlands, do not send that batch as commercial email without prior consent. Autoriteit Persoonsgegevens says Article 11.7 of the Telecommunications Act requires consent for digital direct marketing, including to companies, unless the company published an address for that purpose, the recipient is already a customer and the note is about your own equivalent products, or the recipient is outside the EEA and you follow that country’s rule. A reply the buyer asked for is not that batch. Other countries differ, so the batch is not the default until this check is done.

Do not open by asking for advice and hiding the problem. Politeness without a restatement of the workflow is not a step.

The sentence has to survive without your logo

The demand sentence names one buyer, one costly workflow, and one outcome a colleague could repeat. If it needs your logo to make sense, it is a brand line.

A note that opens with your company and your category fails that test. A note that names one role, one workflow, and asks how they do that work now can earn a reply before anyone knows the name. Problem content has to answer five things: why the break happens, what it costs in time or risk, why the workaround is not enough, what made it urgent, and what to do differently.

Startupik, citing Ali Hajimohamadi, refuses enthusiasm unless the person has the pain, the budget, and the authority to implement. Leaning in only while you explain it live is founder-assisted persuasion. The test is whether they repeat the pitch and still ask. “Better collaboration” fails. Ask in the buyer’s language and room, not on a default US feed. Validate the ICP before outreach, then use the same sentence on a seed homepage. A number you cannot show is not proof.

Proof climbs. A case study is late.

Proof is a climb of buyer behavior, and a public case study is near the top. A logo wall, a raw waitlist, and a kind comment are not rungs.

1. They repeat your sentence.
2. They give time on a call.
3. They share a workflow, a file, or a second person.
4. They accept a bounded manual result and ask again.
5. They introduce a peer.
6. Only then, with permission, a public note a skeptic can check.

Startupik counts calls, shared data, a pilot, a timeline, or a deposit, not likes. A waitlist counts only if a segment takes a stronger step. Startupik’s gate before a deep build is about 20 to 40 good conversations, several strong commitments, and one repeated use case. It is not a weekly quota. Stealth marketing separates curious readers from people who name a use case, and bars fake numbers. Graham lets you consult for one user to learn the mold. Once they pay you for the attention itself, you are a service firm. Keep one workflow, a stop date, and decline custom work the software will not repeat.

One asset, one room, one review date

Run one asset in one room until a review date, and keep it only if someone repeated the sentence and took a step. Otherwise change the list or the line.

The asset is a page with one ask, a manual offer, a short pilot, or a waitlist that records role and use case. Stackmatix says a person owns the path to a meeting. The engine exists when a stranger can book without the founder on every step. Count meetings and pipeline tied to a source, not followers. Log that as an early GTM test.

A second room only redistributes the same asset. Use paid to retarget visitors or to meet a query already in play, not to teach a category from zero. Keep the buyer’s words, change one variable a week, and turn the call into the next piece via content that supports outbound.

Show up where the question already exists

An unknown company borrows a room where the question is already asked. It does not build a media brand and wait.

Answer what buyers ask before a vendor list. Stackmatix uses informational queries, founder posts, and practitioner rooms. If nobody there sees it, it creates no demand. Ask where a forward came from, and write lines an answer engine can quote.

Do not found a room so you have an audience. A building diary fits buyers who want the making, which Adwave says works especially well for developer tools, creator products, and any market where buyers want to see the work being made. In a regulated sale, explain the constraint in private. Stealth still allows the problem, the point of view, and founder expertise. Leave stealth when the reason for it is gone (competitive risk, readiness, or a fundraising moment), or when you are no longer learning and no longer building demand. Ads to an empty page sit with fake numbers.

Repeat the clerical work. Keep the judgment manual.

Automate the rerun of a loop you understand. Leave the buyer, the sentence, and anything a prospect will read with a person until replies exist.

Outbound starts on a spreadsheet and a clean domain, and adds a sequencer only after personal notes earn replies. Graham does the job by hand, then automates the bottleneck he has felt. A manual first version is safer than automation that solves nobody’s problem. After that, software may log the call, cluster objections, draft from one approved insight, remind you, and update the CRM once a person records the outcome. You still pick the wedge, the kill, and the send. Chasing every objection overfits the loudest user, which Startupik flags.

Do not start with a brand program, a launch, or a crowd

Do not open with a brand program, a launch week, or a crowd. Those skip the only demand an unknown company can count: one buyer taking a step.

Leave these in the drawer:

• Identity work before the sentence survives a conversation.
• A launch across several publications.
• A big partner as the source of the first users.
• A waitlist with no segment, or a perk instead of a use case.
• A community you must fill, or a diary for other founders.
• Cold awareness, including television bought to look established.
• Staying hidden after the reason is gone, or after you have stopped both learning and building demand.

Teach the workaround when the category has no name, use search when the query exists, and talk privately when the data cannot be shown. Graham says an idea with no way to recruit users by hand is probably a bad idea, at least for those founders.

I would build the demand loop before the name

I would not start by making the company look known. I would write the buyer and the sentence, run one motion until it produces a commitment or a clear no, and leave the repeated notes as a workflow in your accounts.

I trade as Poldermarketing: one person, remote, in Dutch and English. Hire me for a bounded build or a few days a week. I am an AI-native growth marketer, strong in AI, content, automation, and workflows. Google Ads and Meta Ads are newer for me. I set them up and review them. I do not scale a large media program. I build and execute, not only advise.

The page has one ask. A person follows up until replies exist. Prompts and the log stay in your accounts. The free growth scan shows how the site reads if it still names a category nobody searches. Positioning and messaging makes the sentence repeatable. How I work is an all-in GTM solution for a freshly funded startup anywhere, fully remote, from the first message to the first customers, without a separate specialist for each part.