Channel advice fails startups in a specific way: it arrives as a list. Twelve channels that work in B2B, each with a paragraph, none with a price. But a seed-stage company doesn’t need a list — it needs a way to say no to nine channels and yes to two or three, out loud, on purpose. That’s a formula, not a list.
The Channel-Fit Formula
ICP location × ACV math × founder-hours. Every channel gets scored on all three, and a zero anywhere is a zero overall.
Where is your ICP, actually? Not demographically — physically. Which feeds do they scroll, what do they type into search bars and AI assistants, whose emails do they open, where do they argue about their problems? A channel your buyers don’t inhabit can’t work no matter how well you run it.
What can your ACV afford? Deal size is channel destiny. An ACV under roughly $5K cannot feed outbound math — the hours per closed deal cost more than the deal. Above roughly $25K, self-serve rarely carries the motion, and a founder’s hour per named account becomes a bargain. A $49/month product has about $588 of first-year revenue per customer — there is no room in that number for cold paid acquisition, which is why the paid-ads question at seed is usually answered by arithmetic before it’s answered by opinion.
How many hours exist? Effort is a budget, not a footnote. Every channel has a real weekly cost in hours — outbound wants five to ten, a founder LinkedIn presence three to five, content more — and a channel plan that exceeds the founder’s honest capacity is a plan to fail slowly at everything. This is the factor most channel advice politely ignores, and it’s the one that kills the most plans.
Then the cap: two or three plays, not five. Belkins’ 2026 survey of B2B marketing leaders found teams that excel in 3–5 channels consistently outperform those spread across 10+ — and that’s teams, plural. One more tiebreaker, applied last: channels that compound (search and AI-engine visibility, a founder’s audience, an owned list) outrank channels that rent attention (ads, sponsorships), all else equal — because assets accumulate and rent doesn’t.
The defaults, by motion
The formula usually lands in one of two places, because the sales motion moves every variable at once.
Sales-led at seed (~$10K+ ACV, founder-led sales)
LinkedIn organic — company and founder separately, because trust transfers from a person and warms the outbound. Search + AI-engine visibility as the compounding background bet: bottom-funnel and comparison pages first, not volume blogging. Founder outbound to 50–100 named accounts, worked personally, no tooling. Systematize the word of mouth you already have into structured referral asks, and get two case studies written as outbound ammunition. Refuse at this stage: cold paid, events, community builds.
Product-led at seed (self-serve, ~$50–$200/mo)
SEO/AEO as a core motion — your buyers research before they sign up, and search traffic converts to leads at roughly 3x the rate of paid clicks (First Page Sage, 2025). Customer marketing and referral loops. Lifecycle and activation email as a first-class channel — the launch spike decays, the list is yours forever. Retargeting warm traffic only, capped, with kill criteria written in advance. And if the founder hates posting: authentic presence where the buyers already argue, or nothing — forced founder content reads as exactly what it is.
One channel deserves a special word because almost nobody prices it correctly: AI-engine visibility. When your buyers ask ChatGPT or Perplexity the questions you answer, someone gets cited. The work — answer-first pages, llms.txt, named frameworks, sourced numbers — is classic compounding-channel work, and most incumbents’ content is still tuned for 2019 Google. It’s the rare channel where a seed-stage company can outrank the category leader this year. It runs on the SEO clock, though: months, not weeks.
Write the exit before the entrance
Choosing a channel and knowing when to leave it are the same decision made at different times — and the leaving is easier to decide now, while you’re sober. Every play gets four fields before it starts: what we’re doing, the leading indicator, the number that means it’s working, and the date we decide. Different channels run on very different clocks — paid shows its hand in weeks, SEO takes months to prove anything — and judging a slow channel on a fast clock is how good bets die young. The full verdict system is its own guide.
And before any of this: the formula’s inputs — your real ICP, your ACV, your honest hours — come from looking, not guessing. That’s what an audit is for. Kindling’s free audit runs this exact formula against your motion, your deal size, and your capacity, and hands you the two or three plays it actually supports — with the math shown.