Most launch advice is written by people selling launch services, or by consumer founders whose playbook does not survive contact with a six-week B2B sales cycle. This is neither. Twenty short chapters on what actually moves a B2B launch, what quietly wastes a week, and how to read the aftermath without flattering yourself. Read it front to back, or jump to the part you are stuck on.
1. What a launch actually is
A launch is not an event. It is the moment you start telling a specific group of people that something specific is now possible.
Founders treat launch as a date. It is better understood as a transition: the point where you stop asking people what they would want and start asking whether they want the thing you built. Everything before it is research. Everything after it is evidence.
That reframing matters because it changes what success looks like. If a launch is an event, success is attention. If a launch is a transition, success is learning whether the thing is wanted, as fast as possible, from the right people. Those two definitions lead to completely different plans.
The practical consequence: a launch that reaches two hundred perfectly-matched people and converts eight of them is a better launch than one that reaches fifty thousand strangers and converts thirty. The second one feels better and teaches you less.
2. The launch-day myth
Almost no B2B company traces its trajectory to a single day.
The myth says there is a day, and on that day a wave arrives, and afterwards you are a company. The reality for most B2B products is a slow accumulation: a handful of users from a community post, a few from direct outreach, one from a conversation six months earlier that finally landed.
This matters because the myth causes two expensive mistakes. The first is over-investing in the day: spending three weeks on assets for a spike that lasts eleven hours. The second is reading the day as a verdict: concluding from a quiet Tuesday that the product is dead, when the truth is that the wrong four hundred people saw it.
What to do instead: plan a launch period, not a launch day. Two to four weeks of deliberately telling different audiences, in sequence, so that a poor result in one channel does not read as a verdict on the product.
3. What must be true before you launch
A short list. Shorter than most founders want it to be, and less negotiable.
The product must solve one problem end to end without you intervening. Not every problem. One. If a user has to email you for the thing to work, you do not have a product yet, you have a service with a login screen.
The landing page must name the buyer and the problem in the first sentence. Not the technology, not the category, not the vision. If a stranger cannot tell within five seconds whether this is for them, every visitor you send is wasted.
You must have a way to be reached and to respond within hours on launch day. Not a ticketing system. A monitored inbox. Early users forgive bugs and do not forgive silence.
You must have error monitoring and basic funnel analytics already running. Launching without instrumentation means the most information-dense week of your product's life passes unrecorded.
What is not on the list: a pricing page for every segment, a full onboarding flow, integrations, a mobile app, a brand refresh, or a demo video. All of those can wait, and most of them are procrastination wearing a project plan.
4. Positioning before pixels
The most common cause of a quiet launch is not a bad product. It is a description nobody can act on.
Positioning is the answer to three questions, in writing, before you design anything: who is this for, what does it replace, and why now. If you cannot answer all three in one sentence each, the launch will underperform regardless of channel strategy.
"What does it replace" is the one founders skip, and it is the one buyers care about most. Nobody adopts in a vacuum. They are currently solving this problem with a spreadsheet, a contractor, a competitor, or by ignoring it. Your launch message has to speak to whichever of those they are doing.
A useful test: write the sentence a happy customer would use to describe you to a colleague. If that sentence needs a paragraph of setup, your positioning is not ready. If it sounds like a category you invented, it is definitely not ready.
This is the work Cafiyn Lens exists to do, but the guide works whether or not you use a tool for it. What matters is that the answers are written down and specific before anything is designed.
5. Where your first users actually are
They are almost never in the place you were planning to post.
Your first twenty users come from one of four places, roughly in order of likelihood: people you already know who have the problem, communities you have participated in for months, direct outreach to people who have publicly described the problem, and an existing audience if you happen to have built one.
Notice what is absent from that list: broad social posts, paid ads, and press. Those can amplify a launch that is already working. They do not start one.
The uncomfortable exercise: before launch, write down the actual names of fifty people who have the problem you solve. Not segments. Names. If you cannot get to fifty, you do not have a distribution problem, you have a research problem, and launching will not fix it.
6. Launch channels, ranked by what they actually return
Effort and ceiling are not correlated. Some of the cheapest channels have the highest upside.
| Channel | Effort | Ceiling | When it is worth it |
|---|---|---|---|
| Your own list | Low | Medium | Always first. People who already opted in convert at rates no cold channel touches. If the list is empty, that is the real finding. |
| Communities you are in | Low | Medium | Places you have posted for months before launch day. Showing up only to promote is the fastest way to get removed. |
| Direct outreach | Medium | High | The most reliable B2B launch channel and the least glamorous. Fifty well-researched emails beat one thousand impressions. |
| LinkedIn / X | Medium | Medium | Works if you have built an audience. Does close to nothing on a cold account, regardless of how good the post is. |
| Product Hunt | High | Medium | A traffic spike, not a customer channel. Good for social proof and backlinks. See the chapter below before committing a week to it. |
| Hacker News | Low | Very high | Highest variance of any channel. Free to try, brutal feedback, occasionally transformative. Works for developer tools, rarely for anything else. |
| Paid ads | Medium | Low at launch | Almost never right at launch. You are paying to learn things a hundred conversations would have told you for free. |
| Press / PR | Very high | Low | Rarely worth it for an unknown B2B product. Journalists cover funding, people, and conflict, not feature launches. |
The pattern worth noticing: the two highest-ceiling channels for B2B are direct outreach and Hacker News, and they are both low-cost. The two most effort-intensive are Product Hunt and press, and both have modest ceilings. Founders systematically invert this, because the high-effort channels feel more like launching.
7. Product Hunt, honestly
A useful marketing asset. A poor acquisition channel. Plan accordingly.
Product Hunt reliably delivers a one-day traffic spike, a permanent backlink, a badge you can put on your site, and a burst of feedback from people who look at new products professionally. Those are real benefits and they are worth something.
What it does not reliably deliver is customers who are still there in a month. The audience is largely other builders, and builders are curious rather than in-market. For a developer tool that overlap is genuine. For most vertical B2B software it is not.
If you do it: the outcome is decided by whether you can get a meaningful number of people to show up on one specific morning. That is a function of your network, not your product. If you do not have that network, a lower rank is not a judgement on what you built.
The honest cost: doing it properly consumes most of a week. Weigh that against fifty researched outreach emails, which for most B2B products will produce more revenue and far better information.
8. Hacker News, honestly
The highest-variance channel available to you, and close to free to attempt.
A Show HN post costs fifteen minutes. It will most likely do nothing. Occasionally it changes the trajectory of a company. That asymmetry makes it worth attempting for almost any technical product, provided you can handle the feedback.
The feedback is the part people underestimate. Hacker News comments are direct to the point of bluntness, and they will find the weakest assumption in your product within the hour. That is genuinely valuable and genuinely unpleasant. Read it, extract the two or three substantive critiques, and discard the rest.
What tends to land: a plain title, a real technical detail, a working demo that needs no signup, and a founder answering every comment personally and without defensiveness.
What tends not to: marketing language, a signup wall before anyone can see anything, and arguing with critics.
9. Reddit, Slack, and Discord
High-trust, high-intent, and unforgiving of anyone who shows up only to promote.
Niche communities are where B2B buyers describe their problems in public, which makes them the single best research surface available and a decent launch surface if you have earned standing there.
"Earned standing" is the whole condition. Communities can tell instantly whether you have been present for months or arrived this morning with a link. The first gets a conversation. The second gets removed, and sometimes gets a reputation.
The workable approach: participate for a few months before you need anything, answering questions in your area without mentioning what you are building. When you launch, post it as a thing you made because you had the problem, not as an announcement. Then reply to everyone.
11. The list you own
The only distribution channel no platform can take away from you.
Every other channel here is rented. Algorithms change, subreddits close, communities die. A list of people who gave you their email because they wanted to hear about this is the one asset that persists, and it consistently outperforms everything else on conversion.
If the list is empty at launch: that is not a launch problem to solve this week, it is a finding. It means you built before you gathered demand. The fix is not a better launch email, it is a month of conversations with people who have the problem.
If the list exists: the launch email should be short, name the problem in the first line, link to one thing, and come from a person rather than a brand. Reply rate matters more than open rate. A reply is a conversation; an open is a reflex.
12. Press, and why it rarely works
Journalists cover funding, people, and conflict. A feature launch is none of those.
Founders routinely spend weeks on press outreach for an unknown B2B product and get nothing, then conclude they wrote a bad pitch. Usually the pitch was fine and the premise was wrong: there was no story, only an announcement.
Trade press in your specific vertical is a different matter and is often genuinely reachable, because their readers actually want to know that a new tool exists in their niche. That is a much better use of the same hours than pitching general technology publications.
When press does work: you raised money, you have an unusual founding story, you have data nobody else has, or you are taking a public position on something contested in your industry. Absent one of those, deprioritise it.
13. The assets you actually need
Four. Everything beyond them is optional and most of it is delay.
A landing page that names the buyer and the problem above the fold, shows the product doing the thing, and has one action. Not three. One.
A short demo, ideally under ninety seconds, that shows the actual product rather than an animated abstraction. Screen recordings outperform produced videos for B2B, because they are evidence rather than marketing.
One paragraph of copy you can paste into a community post, an email, and a social post without rewriting it each time. Consistency here is worth more than bespoke copy per channel.
A way to pay or join. If you are not charging yet, a waitlist with a single field. Anything longer costs you signups for information you can ask for later.
Not needed on day one: a press kit, a brand guideline, testimonials you do not have, a comparison page, or a pricing calculator.
14. Deploy readiness
The goal is not zero failures. It is knowing about failures before your users tell you.
Before you send traffic, four things should be true. Error tracking is live and you have confirmed it fires by deliberately triggering an error. You can roll back in minutes without a deploy pipeline that takes twenty. Someone is watching during the hours your traffic will actually arrive. The signup path has been walked end to end, on a fresh account, on a phone, in the last twenty-four hours.
That last one catches an embarrassing number of launch-day failures. The path you test constantly as a logged-in developer is not the path a first-time user takes.
What not to do: ship a large change the morning of the launch. The temptation is enormous and the failure mode is a broken signup flow during your only spike.
15. What to watch while it is live
Five surfaces, checked in a loop, for the hours that matter.
Errors. New exception types matter far more than volume. A spike in a known error is load. A new error is a broken path.
The funnel. Specifically the step-to-step drop. If visitors are arriving and not starting signup, the problem is the page. If they start and do not finish, the problem is the form or the pricing.
First value. How many people reached the moment the product is actually useful. This is the only number on launch day that predicts anything about next month.
Inbound. Email, social replies, community comments. Confused users mostly do not write in, they leave, so each message represents a larger silent group.
Infrastructure. Response times and database load, but only enough to know whether slowness is causing the drop-off you are seeing elsewhere.
16. Real signal versus flattering numbers
Every vanity metric has a real counterpart that is harder to look at and worth more.
| What feels good | What actually tells you something | Why |
|---|---|---|
| Page views on launch day | Signups from people who match your ICP | Traffic without fit is a bill, not a signal. A hundred wrong-fit visitors teach you nothing. |
| Product Hunt rank | How many upvoters activated a week later | Rank measures your ability to mobilise a network on one day, not whether the product is wanted. |
| Total signups | Signups that reached first value | A signup is an intention. Activation is evidence. Only one of them predicts revenue. |
| Social impressions | Replies that describe a real problem | One person explaining their pain in detail is worth more than ten thousand silent impressions. |
| Newsletter subscriber count | Open and reply rate on the launch email | A list that does not open is a list that does not exist. |
| Uptime on launch day | Time to first response when it broke | Everyone breaks at some point. Buyers remember how fast you answered, not that it was perfect. |
The one number to hold onto: how many people who match your ICP reached first value and came back once without being prompted. It is a small number in the first week. It is also the only one that has ever predicted whether a product works.
17. The support spike
Launch week generates more support volume per user than any period that follows. That is an opportunity, not a cost.
Every confused message is a description of a gap between what you built and what someone expected. In normal operation you would pay a researcher for that. During launch week it arrives free and in volume.
Answer everything personally, fast, and in a human register. Not because it scales, but because it will not need to for long, and because early users who receive a real reply from a founder become the people who tell others.
Keep a running tally of question types. Any question asked three or more times is not a support issue, it is a product or copy defect, and it should be fixed in the interface rather than answered again.
18. Turning feedback into a roadmap
Launch feedback arrives in a volume that feels like a mandate. Most of it is not.
Separate every piece of feedback into three buckets. Confusion means the product works but the explanation does not: fix the copy, which is cheap and immediate. Missing capability means someone wanted a thing that is not there: only count it if the person matches your ICP. Wrong fit means they wanted a different product: this is the most valuable and most commonly mishandled bucket.
The trap is building for the wrong-fit bucket, because it is loud and the requests are concrete. Feedback from people who will never buy is not a roadmap, it is a distraction wearing the costume of validation.
The filter that works: for every request, ask whether the person asking has the problem you set out to solve. If not, log it and move on. If yes, and three of them ask, it is real.
19. The week after
Traffic collapses. This is normal, universal, and not information.
Every launch spike decays within days. Founders reliably experience this as failure. It is not: it is what a spike is. The information is in what is left behind once the spike clears.
The three questions for week two: did anyone who matched the ICP reach first value? Did any of them come back without being prompted? Did anyone pay, or say they would if a specific thing existed?
Three yeses means you have something and the job is distribution. One or two yeses means the product is close and the positioning or the audience is off. Zero yeses, from an audience that genuinely matched, is the real signal, and it is worth more than a comfortable maybe. Most founders never get a clean answer because they never showed it to a clean audience.
20. Relaunching without shame
Launching once is a convention, not a rule. Most durable B2B products launched several times.
Nobody is keeping score. The audience that saw your first launch is overwhelmingly not the audience that will see the second, and the fraction that saw both will not remember. The reluctance to launch again is entirely internal.
Legitimate reasons to launch again: a new segment you can now serve, a capability that changes who the product is for, a repositioning that makes the value obvious, or simply that the first attempt reached the wrong people.
What to change: the audience first, the positioning second, the product last. Founders reliably reverse that order, rebuilding for months before ever testing whether a different audience with the same product would have said yes.
FAQ
How long should a B2B launch take to prepare?
Two to four weeks of preparation is typical once the product is genuinely usable. Most of that time goes into positioning, assembling a list of people to tell, and making the first-run experience survivable. If preparation is stretching past six weeks, the delay is usually a confidence problem rather than a readiness problem.
Should I launch on Product Hunt?
Only if you already have a network you can mobilise on a single day, and only if you value the backlink and social proof more than the customers. Product Hunt reliably produces a traffic spike and unreliably produces retained users. It is a marketing asset, not an acquisition channel.
What is the minimum I need before launching?
A product that solves one problem end to end without a human intervening, a landing page that names the buyer and the problem in the first sentence, a way to collect payment or a waitlist, error monitoring, and a way for a confused user to reach you within a few hours. Everything else is optional on day one.
How many people should I tell before launch day?
Every single person who has ever expressed the problem you solve. If that list is shorter than fifty people, the launch is premature: you have not talked to enough of the market to know whether the product is wanted.
What does a failed launch actually look like?
Not silence. Silence usually means the wrong audience saw it. A genuinely failed launch is when the right audience sees it, understands it, and still does not want it. Those two failures have completely different fixes, and telling them apart is the whole job of launch week.
Can I launch more than once?
Yes, and most successful products do. Launches are not a one-time event; they are a repeatable act of telling a specific audience that something specific is now possible. A new segment, a major capability, or a repositioning all justify launching again.
How soon should I know whether it worked?
Traffic tells you within a day. Fit takes two to four weeks, because you need enough people to reach first value and then decide whether to come back. Judging a launch on day-two numbers is the single most common way founders talk themselves into abandoning something that was working.
What channels should a B2B product launch on first?
The channels where your buyer already complains about the problem. For most B2B products that means a founder-led post on LinkedIn, a direct email to the people who expressed the problem, and a presence in the two or three communities your buyers actually read. Paid ads, Product Hunt, and press are amplifiers, not a first channel: they work once you have a message that converts warm traffic.
How much should I spend on a launch?
At launch, close to nothing on paid acquisition. The scarce resource is not budget, it is a message that lands and a product that delivers first value quickly. Spend your effort on positioning, on the first-run experience, and on personally reaching the people most likely to care. Add paid spend only after warm traffic converts, so you are amplifying something that works rather than paying to discover it does not.
Should I gate the product behind a waitlist or launch open?
A waitlist is worth it only when you genuinely cannot serve everyone at once, or when scarcity is part of the positioning. Otherwise it adds friction between interest and first value, which is exactly where launches leak. If the product is ready and can absorb the traffic, launch open and let people reach value the same day they hear about you.
Where this connects to Cafiyn
Two chapters of this guide are really about knowing your buyer before you announce anything: positioning and where your first users are. That is the work Cafiyn Lens does, and getting it wrong is the most common reason a technically good launch stays quiet.
The chapter on reaching those people is where Cafiyn FlyWheel picks up: turning a defined buyer into actual conversations, at a volume a founder cannot sustain by hand.
For the evidence behind the channel rankings in this guide, the startup case study library tears down how fourteen companies actually got their first customers, with every source linked. Not one of them traces its first customers to press coverage.
10. LinkedIn and X
Amplifiers of an audience you already have. Not a way to create one on launch day.
If you have spent a year posting usefully about your problem space, a launch post will do real work. If you have not, it will reach your existing connections and stop, no matter how well written it is. This is a distribution reality, not a copywriting problem, and no amount of editing changes it.
The one exception worth knowing: a specific, concrete story about a problem you solved tends to travel further than an announcement that a product exists. "We kept losing deals at the security review, so we built X" outperforms "Introducing X" by a wide margin, because the first sentence is about the reader and the second is about you.
If you are launching in six months, the highest-return action available today is to start posting now, so the channel exists when you need it.