Before product-market fit, growth work has one job: find a single channel that reliably delivers customers you can learn from, not one that delivers volume. Spending real money to scale an unproven channel just burns cash faster while masking whether anyone actually wants what you built.
Quick answer: Pre-PMF, run the
Bullseye Framework— brainstorm every plausible channel, cheaply test the top few, then focus once one proves repeatable. Hold off on paid scaling until retention holds, not just signups, or you'll pay good money to hide a leaky bucket.
What "Growth" Actually Means Before Product-Market Fit
Pre-PMF growth means running small, cheap distribution experiments to learn which channel and which message a real segment responds to — not maximizing signups or revenue. The goal is a repeatable acquisition motion you understand, not a large one you can't yet explain.
Founders arriving from a post-PMF mental model default to "more" — more spend, more channels, more surface area. That instinct runs backwards before you've found fit.
Distribution pre-PMF is a discovery tool, not a growth lever — every dollar and every week spent testing a channel should answer a question, not just move a number.
This is the same shift we describe in our piece on discovery over delivery before product-market fit: the job isn't to ship and sell as much as possible, it's to run experiments that each produce a confident yes or no.
Growth channels are just another surface for that discovery — arguably a more honest one, since a market's real response to your acquisition message is harder to fake than a customer interview. If you're the one holding every function right now, our founder-PM complete guide maps out where distribution testing fits alongside the rest of the job.
Two failure modes bookend this stage:
- Chasing scale too early. Pouring budget into paid ads or a big launch before you know what converts burns cash on a channel you haven't validated and can't yet explain.
- Avoiding distribution entirely. Waiting for the product to feel "ready" before testing any channel starves you of real-world signal that discovery interviews alone can't produce.
Both look like discipline. Neither is. Founders who navigate this well treat growth as a parallel discovery track, run at a pace and budget that match how unproven the product still is.
The Bullseye Framework: How to Test Channels Without Betting the Company
The Bullseye Framework, developed by Gabriel Weinberg and Justin Mares in their book Traction, gives you a structured way to test distribution channels cheaply before committing real budget to any one of them. It works in three rings: brainstorm broadly, rank by running small real tests, then focus once a channel repeats.
Weinberg and Mares catalog 19 traction channels — from viral loops and SEO to sales, business development, and community building. They argue most founders instinctively gravitate to two or three based on what they've seen work elsewhere, ignoring the rest. The framework's entire point is forcing a wider first pass before narrowing.
Ring One: Brainstorm Every Plausible Channel
List every channel that could conceivably work, even ones that feel unlikely or uncomfortable. Common categories worth forcing yourself through:
- Viral and referral loops — does usage naturally expose the product to new users?
- Content and SEO — is there a search intent your product answers?
- Paid social and search ads — is there a keyword or audience cheap enough to test?
- Sales and business development — does your buyer respond to a direct conversation?
- Community and existing platforms — is there a forum, marketplace, or platform your users already live on?
- PR and unconventional PR — is there a story here a journalist or niche newsletter would run?
For each, generate three to five concrete tactics, not just the category name. "Content marketing" isn't a test; "publish three comparison posts targeting a specific long-tail query" is.
Ring Two: Rank and Run Cheap Real Tests
Sort your brainstormed list into three buckets: promising, possible, and long-shot. Then run genuine, small-scale tests on your top three to five promising channels — not thought experiments, actual attempts with real (small) budget or time.
The output you want from each test is concrete: an estimated cost per acquired customer, a rough ceiling on how many customers that channel could plausibly supply, and how hard it was to execute. A channel that's cheap but has a ceiling of fifty customers total isn't a growth channel — it's a nice-to-have.
Ring Three: Focus Once One Channel Repeats
Once a channel shows a repeatable pattern — the same tactic producing customers a second and third time at a similar cost — go all-in on it. Weinberg and Mares are explicit that most successful companies win through one dominant channel, not five mediocre ones running in parallel.
| Ring | Core Question | Typical Duration | Exit Criteria |
|---|---|---|---|
| Brainstorm | What could conceivably work? | 1-2 days | 15-19 channels listed with 3-5 tactics each |
| Rank | What actually works, cheaply? | 2-6 weeks | 3-5 channels tested; cost and ceiling estimated per channel |
| Focus | What's repeatable? | Ongoing | One channel produces customers a second and third time at similar cost |
A founder-PM running this well doesn't skip Ring One to save time — skipping it is exactly how you end up over-invested in the channel you were already biased toward, rather than the one that's actually working. That's the same founder-as-HiPPO dynamic worth watching for in growth decisions specifically: the highest-paid person's opinion tends to favor whichever channel they personally find validating, not necessarily the one the cost-per-customer data supports.
Why Premature Paid Acquisition Hides Your Retention Problem
Paid acquisition before product-market fit is dangerous because it manufactures top-of-funnel numbers that look like growth while doing nothing to prove people stick around. Spend enough on ads and signups will rise regardless of whether the product solves a real problem — which is exactly why paid spend is the wrong tool for pre-PMF discovery.
Eric Ries's Lean Startup framing of the three engines of growth — paid, viral, and sticky — is useful here because it puts sticky (retention) first, as a precondition rather than an afterthought. You can't responsibly judge a paid or viral engine's economics until you know your retention curve, because CAC math depends entirely on how long a customer sticks around and what they're worth over that time.
The Leaky Bucket Nobody Notices
Picture two founders, both spending $2,000 on ads and getting 200 signups. One product retains 35% of those users past week four; the other retains 4%. The top-of-funnel report — cost per signup, click-through rate, even day-one activation — can look identical for both.
That's the trap: paid acquisition metrics are almost entirely blind to retention, and retention is the one number that actually indicates product-market fit. Spending to scale acquisition before that number is healthy just means paying to refill a bucket with a hole in the bottom, faster.
Paid spend is a magnifying glass, not a fix: it makes a working channel look better and a broken product look identical to a working one — for a while.
Sean Ellis's product-market fit survey asks a simple question: how would users feel if they could no longer use the product? He found that companies clearing roughly 40% "very disappointed" responses were the ones that went on to scale successfully. Below that threshold, more paid spend just buys a bigger group of people who wouldn't miss you.
Brian Balfour's Four Fits framework (Reforge) adds a related angle: a channel only works long-term if it has channel-model fit, meaning its acquisition cost has to make sense against your margins. Mapping what actually happens between a user's first click and the moment they'd genuinely miss the product — the territory covered in our customer journey guide — is what tells you whether a drop-off is a fixable onboarding problem or a fundamental no-one-wants-this problem.
Watch for these tells that you're paying to mask a retention problem rather than fund real growth:
- Signups are rising but week-four or month-one retention is flat or falling.
CAClooks fine, butLTVis still a guess, not a measured cohort number.- The team's growth conversation is about spend and click-through rate, never the retention curve.
- You've never run a formal
PMFsurvey or cohort retention analysis, but you're already running paid campaigns.
Traction Experiments vs. Scaling Experiments: Different Questions, Different Budgets
A traction experiment asks whether a channel can work at all; a scaling experiment asks how much you can responsibly pour into a channel you already know works. Confusing the two is how founders end up scaling an unproven channel, or endlessly "testing" a channel that already proved itself weeks ago.
| Dimension | Traction Experiment | Scaling Experiment |
|---|---|---|
| Core question | Does this channel produce customers at all? | How much can we pour in before returns drop? |
| Budget | Small, capped, treated as a learning cost | Larger, tied to a known CAC:LTV ratio |
| Primary metric | Cost per customer, channel ceiling | Marginal CAC, payback period, retention by cohort |
| Duration | Days to a few weeks | Ongoing, reviewed monthly or quarterly |
| Success looks like | A repeatable pattern at small volume | Same or better economics at 5-10x volume |
| Failure looks like | No repeatable pattern after honest tries | Rising CAC or falling retention as volume grows |
Signals a Channel Is Ready to Graduate From Test to Scale
A channel is ready to scale, not just test, when three things are true at once:
- You've acquired customers through it more than once, at a similar cost, without a one-off stroke of luck (a press hit, a viral post) driving the number.
- Retention among those customers matches or beats your other channels — a channel that brings in users who churn faster than average is quietly worse than it looks on a
CACdashboard alone. - You can articulate why it works — which segment, which message, which moment in their day — well enough that a new hire could roughly repeat it without you.
Every channel test is really a Jobs-to-be-Done experiment wearing a distribution costume: the copy, targeting, and channel context are hypotheses about which job a segment is hiring your product for. Our Jobs-to-be-Done complete guide covers how to read a channel's response — high engagement but low conversion, say — as a signal about which job you nailed and which one you didn't.
If any of the three conditions above is missing, you're still running a traction experiment, even if you've technically been spending money on the channel for months. Time spent is not the same as evidence gathered.
Building the Distribution Habit Into Your Founding-PM Workflow
Growth discipline pre-PMF isn't a one-time framework exercise — it's a habit of running small tests, writing down what you learned, and revisiting your channel ranking every few weeks instead of emotionally committing to whichever one got attention last. If you're early in the role, our guide to a founding PM's first 90 days suggests a rough cadence — one Bullseye pass every four to six weeks — as a forcing function instead of open-ended exploration.
The habit that actually compounds is capturing, close to when it happens:
- The message that landed and the one that didn't.
- The segment that responded versus the one that went quiet.
- The cost that surprised you, whether higher or lower than expected.
- Whatever made you rethink the channel ranking mid-test.
Neither replaces running the tests yourself. What they're designed to do is make each test's learning durable enough to inform the next Bullseye pass, instead of evaporating the moment you move on to the next fire.
Key Takeaways
- Pre-PMF growth is a discovery tool, not a scale lever — every channel test should answer a question about fit, not just move a vanity metric.
- The Bullseye Framework (Weinberg & Mares, Traction) forces you through three rings — brainstorm every plausible channel, cheaply rank the top few, then focus once one repeats — instead of defaulting to the channel you already like.
- Paid acquisition before product-market fit can hide a retention problem, since rising signups can mask a flat or falling retention curve; Sean Ellis's roughly 40% "very disappointed" benchmark is a useful gut-check before scaling spend.
- A traction experiment and a scaling experiment ask different questions — "does this work at all" versus "how much can we responsibly pour in" — and mixing them up leads to either premature scaling or endless, evidence-free testing.
- A channel is ready to scale only when it's repeatable, retains as well as your other channels, and you can explain why it works well enough for someone else to repeat it.
- Treat every channel test as a Jobs-to-be-Done experiment: the response tells you as much about which customer job you nailed as it does about the channel itself.
- Capture what you learn close to the moment you learn it — a growth habit that compounds depends on notes taken during the test, not reconstructed weeks later.
Frequently Asked Questions
How much should a pre-seed startup spend on growth before product-market fit?
Enough to run real, small-scale tests on three to five channels — typically a few hundred to a few thousand dollars per channel, not a full marketing budget. The goal is learning cost per customer and channel ceiling, not volume, so cap spend at whatever it takes to get an honest read.
Is paid advertising ever appropriate before product-market fit?
Yes, but only as a capped, short traction test, not a scaling motion. Use it to learn whether a specific message and audience convert at all, and hold off on meaningfully increasing budget until retention among those paid users matches your other channels.
What is the Bullseye Framework in the Traction book?
It's a three-ring method — brainstorm, rank, focus — for narrowing 19 possible distribution channels down to the one or two that actually work for your product. Gabriel Weinberg and Justin Mares developed it specifically to stop founders from over-indexing on the two or three channels they already know.
How do I know if my channel test failed or if I just haven't tried hard enough?
If you've run three to five concrete tactics on a channel with real, even if small, spend and time, and gotten no repeatable pattern, that's a legitimate no — move to the next promising channel. A single half-hearted attempt with no clear tactic isn't evidence either way.
Should founder-led growth stop once you hire a growth hire or team?
Not immediately. Founder involvement in early channel testing generates the qualitative pattern-recognition — which message, which segment, which moment — that a growth hire needs inherited context to repeat. Handing it off before that pattern is documented usually means relearning it from scratch.