No budget, no brand, no attribution, no team — the honest sequence for finding the one channel that works.
You can't grow what you can't position — start with startup positioning.
0→1 is not a scaled-down version of 1→n. With no budget, brand, or attribution, the paid-acquisition playbook doesn't apply. The whole job is two things:
The proof, honestly: GrantCompass hit 25,000+ users in ~4 months on $0 ads / $0 PR. But that result was measured on a live site, not cleanly caused — 76 of 87 experiments were confounded by a platform-wide wave. Copy the sequence, not the number.
Zero to one, in growth. When you have no budget, you can't buy growth yet. Most startups spray seven channels — GEO and AI search, SEO and content, community, founder-led social, product-led loops, partnerships, and cold outreach. The move is to find the one channel that works, then double down. For GrantCompass, that channel was GEO and AI search. Then it compounds: zero to 25,000 users in about four months, on $0 of ads — measured on a live site, not caused.
Going from zero to your first real cohort of users is a different job from growing an already-working machine — and most people who've only done the second one fail at the first. At 0→1 you have no budget to buy your way in, no brand that makes strangers trust you, no attribution to tell you what's working, and no team to parallelize. The polished paid-acquisition playbook — bid strategies, MMM, incremental lift tests — assumes a working funnel and money to feed it. You have neither.
So the job shrinks to something almost embarrassingly simple to state and brutally hard to do: find one channel that works and build a product people come back to. Everything else — the dashboards, the second channel, the growth team — is a reward you earn by getting those two right first. And you can't even start until you know exactly who the product is a must-have for, which is a positioning problem before it's a growth problem.
Traction is a ladder, not a spray. Each rung only holds weight once the one below it does — so the fastest path is almost always slower than it feels, because it refuses to skip a step. Spraying five channels at a pre-retention product just distributes your failure across more surfaces.
Read it bottom-left to top-right. A product no one talks about (rung 1 missing) makes every channel expensive and every user a one-time visit. Find one channel before you build the loop, because the loop is channel-specific — a referral loop, a content loop, and an integration loop are different machines. And you only earn the right to scale (paid, headcount, a second channel) once the first three rungs bear weight. Most startups die trying to climb rung 4 while rung 1 is still broken.
Peter Thiel's blunt claim in Zero to One is that most startups get zero distribution channels working — not too few — and that a single channel done well beats a spread of half-committed ones. Gabriel Weinberg & Justin Mares' Traction agrees on method: their Bullseye framework says brainstorm all 19 channels, cheaply test a promising few, then pour everything into the one that's actually moving. At 0→1 your scarcest resource is attention, not options.
The channel that built GrantCompass. One well-structured answer keeps getting cited for months. Slow to start, near-zero marginal cost, brutal moat once it takes.
Same compounding logic as GEO with a longer time-to-signal. The asset you wish you'd started 12 months ago — which means the best time to start is now.
Highest-trust, lowest-scale. Doesn't compound on its own — you have to keep showing up — but it's how you win your first 100 true users and learn their exact words.
Distribution you fully control and the fastest time-to-signal on this list. Compounds while you post, stalls when you stop — it's a habit, not an asset.
The best channel if the product creates it — referral, sharing, multiplayer. It compounds beautifully but can't be bolted on afterward; it has to be designed into the product.
Borrow someone else's distribution. Compounds once live, but each deal is bespoke and slow to close, and it only works if you plug into an ecosystem your users already live in.
The unscalable thing you do first. Zero compounding — every user is hand-won — but it's the fastest way to learn who your user actually is and whether they'll pay.
Directional, not a verdict. The fit values are my judgment calls to make the trade-offs visible, not measured data. With all four criteria on, the two compounding channels win — which is the whole lesson. Turn off "does it compound?" and watch community, social, and outreach climb: those are your first-100-users channels, not your first-10,000 ones.
GrantCompass reached 25,000+ users in about four months on $0 ads and $0 PR. It's the clearest evidence I have for this playbook — so it's worth being precise about what did the work and what I can't claim. The sequence was one channel, compounding: a product built to be discoverable → on-site GEO → AI citations → discovery → product-led retention.
The channel was GEO — structured, answer-first pages that AI engines could extract and cite. That produced 192,924 Microsoft Copilot citations in six months (measured in Bing Webmaster Tools), which drove discovery, which sent qualified users to a product built to keep them. No paid acquisition, no PR, no link-building. One channel, done to the point of compounding, feeding a product with retention — exactly the ladder above.
This was measured on a live site, not cleanly caused. 76 of the 87 experiments were confounded — pages ran multiple tests at once, and Copilot's citation ecosystem roughly doubled platform-wide during the window. One page I left untouched was among the top gainers. So the honest framing is: the pages rode and captured a wave as much as they created one. The on-site work positioned them to capture more of it than they'd have gotten by default — but no single tactic gets clean causal credit. Copy the sequence and expect the magnitude to depend on your window and starting authority. The full first-party data — charts, confounders, and all — is in the GrantCompass GEO case study.
Six channels you can run with time instead of money. The two things worth reading closely are the catch — every channel has one — and time-to-signal, because at 0→1 the risk isn't picking a bad channel, it's waiting six months to find out you picked a bad channel.
| Channel | Why it works at 0→1 | The catch | Time-to-signal |
|---|---|---|---|
| GEO / AI-search | You become the cited answer; near-zero marginal cost per new user | Slow to start; partly rides forces you don't control | 4–8 weeks |
| SEO / content | Compounding asset that keeps working after you stop writing | Longest ramp; crowded on commodity topics | 3–6 months |
| Community | Highest trust; you learn your users' exact words and objections | Doesn't compound alone — you have to keep showing up | Days–weeks |
| Founder-led social | Distribution you control; fastest feedback on message-market fit | Stops the moment you stop posting; ties growth to you | Days–weeks |
| Product-led / virality | The product does the acquiring; every user can bring the next | Must be designed in — can't be bolted on later | Weeks (once shipped) |
| Partnerships / integrations | Borrow an installed base; instant reach into the right ecosystem | Each deal is bespoke and slow; you don't own the relationship | 1–3 months |
If you're a founder's first growth hire, the fastest way to lose the room is to launch five channels in month one. The job in 90 days is narrow: prove you can find the one channel and read it honestly. Here's the sequence I'd run.
Talk to real users, sharpen who the product is genuinely a must-have for (Sean Ellis' "how would you feel if you could no longer use this?" survey is the cleanest read on product-market fit), and wire up the funnel so you can see signup → activation → week-4 retention. No instrumentation = no honesty. You can't pick a channel until you know who you're reaching and whether they come back.
Choose a single zero-budget channel that can plausibly reach your ICP and compound, and commit. Before you pour effort in, build the lightweight way you'll read whether it's working — a leading indicator, not a lagging one. Resist the pull to hedge across five channels; a hedged channel is a channel you never really tested.
Read the signal without flinching. Working? Pour in effort and hunt for the loop — the mechanism that makes each new user bring the next. Not working? Kill it without ceremony and move to the next candidate. The bar for day 90 is exactly two things: one channel that works and a product people come back to.
Short answer: after you have retention and one organic channel that works — not before. This is the most expensive sequencing mistake at 0→1, because paid feels like growth while it's really just renting a graph that collapses the day you stop paying.
Turning on ads before you have retention buys you an expensive, temporary spike: users who don't come back, unit economics that only work in a slide, and a growth curve that's really a subscription to a channel. Paid is fuel, not a spark — it makes a fire you already have bigger and faster. So the trigger to switch it on is boring and specific: retention is real, one organic channel is working, and the unit economics survive contact with a spreadsheet. Then paid buys you speed. Before that, it just buys you the illusion of traction while the leak stays open.
You can't grow what you can't position. The sentence you have to finish before any channel.
Read it →How a one-person growth function runs the whole motion — build, measure, iterate — with AI.
Read it →25,000 users, 192,924 citations, $0 ads — and the honest read on what it proves.
Read the case study →The GEO field guide — how the compounding channel from this piece actually works.
Read the field guide →If you're seed–Series A and looking for someone to find your one channel and build the loop — with no budget, no team, and no vanity metrics — let's talk.
Last updated July 2026 · Part of an in-progress series on early-stage growth & AI-native marketing.