Product Launches
First make sure it actually failed, and where. Then decide what to change before you spend another dollar.
By Shannon Kearns, published 2026-10-04, 12 min read.
The work after a failed launch goes in order: diagnosis, then a decision about what to change, then (maybe) a relaunch.
When a product launch fails, the most expensive mistake is fixing the wrong thing. The numbers come in soft, leadership wants a plan by Friday, and the team reaches for whatever is easiest to change: new ads, a new headline, a bigger push.
Those fixes might feel like action, but they're usually aimed at the last step in the chain. Usually where a launch goes wrong is further up, in a decision about who the product was for, what it had to prove, or whether it was ready for more people.
A launch failed when it missed the success definition your team agreed to before launch. If you didn't agree on one, you can't yet tell a failed launch from a slow start, and that's the first thing to fix.
This sounds picky, but it matters. I've watched teams declare a launch dead at day 21 when the sales cycle was 60 days, and I've watched others call a launch fine for a full quarter because signups looked great while almost nobody came back and used the product twice.
If you don't have a success definition, write one now, honestly, with what you know today. Set it by funnel stage so you can see where the drop happens, not just that it happened.
| Funnel stage | What to look at | A problem here usually points to |
|---|---|---|
| Awareness | Traffic from target accounts, ad/social impressions and engagements, press metrics, email opens | Channel or audience choice |
| Consideration | Demo requests, trial starts, time on key pages | Messaging or positioning |
| Conversion | Trial-to-paid rate, win rate, sales cycle length | Offer, pricing, or ICP fit |
| Activation | Share of new users who reach the first moment of value | Onboarding or product readiness |
| Retention | Repeat usage, early churn, expansion | Fit with this segment |
Read the table from the top down and stop at the first stage that's clearly below where it should be. That's where you start digging. Everything below it is downstream of the real problem.
(If your problem is at activation or retention and your team is debating ad creative, that's a strong sign you're about to fix the wrong thing.)
Three checks before you call it:
If a stage is well below target and not improving after a full cycle, the launch failed at that stage. Now you know where, and the next job is to stop making it worse.
Pause anything that sends more people into the broken stage, keep everything that doesn't depend on it, and set a date for the diagnosis. Do this in the first week, before you know the full cause.
The instinct to either push harder or pull everything makes sense. But pushing harder buys more traffic into a leak, and pulling everything wastes the prep and sends a panicked signal to your team and your customers.
What I do when I see signs the launch isn't going to plan:
You find the cause by collecting evidence from three places: the funnel data, the customers who didn't buy or didn't stick, and the people inside the company who heard it first. Data tells you where people dropped. The conversations tell you why.
The customers. Talk to four to six people in each group:
Ask about what happened, not about your product. "Walk me through what you were trying to do when you signed up" gets you more than "what did you think of onboarding?" Record every call.
The internal team. Talk to sales, CS, support and product one at a time. Ask each person:
That third question is uncomfortable, and it's the one that pays off. Someone usually saw part of it coming.
If the answers to the first question don't match, that's a finding on its own. Only about 19% of people inside a company give the same answer about what the product's value is.
Then match what you're seeing to where it usually traces back:
| What you're seeing | Where it usually traces back |
|---|---|
| Traffic came, nobody converted | Messaging or positioning. Buyers didn't see their problem in what you said. |
| Sales went back to the old pitch within weeks | Messaging written from the inside out, or no real alignment with sales before launch |
| Strong signups, almost nobody came back | Product readiness, or weak fit with this segment |
| Wins from one segment, losses everywhere else | An ICP that's too broad |
| Nobody agrees whether it worked | Success metrics weren't set before the plan |
| Prospects keep saying "so it's like [competitor]" | Positioning that isn't differentiated |
Treat each match as a hypothesis, not a verdict. Product marketer Irina Nica made this point in a Sharebird answer on launch post-mortems: "conversion dropped" is an observation, and the useful move is to name the likely reason and assign someone to go confirm it. (I went through each of these areas, with warning signs, in Why Product Launches Underperform.) If you're early-stage, I broke down why product launches fail before launch day, and the cheap tests that catch it.
A product launch post-mortem is one working session where the team compares what it expected with what happened, explains the gap, and decides what to keep and what to change. The most reliable structure is the U.S. Army's After Action Review, which Wharton describes as four questions:
The order matters. Agreeing on facts before anyone explains them keeps the meeting from turning into competing stories.
How I run it:
Share the output with the whole team, including the parts that aren't about their function. The post-mortem gives you a root cause. The next section is the decision it feeds.
Relaunch only when the thing that caused the failure has actually changed. If you relaunch with the same audience, the same claim and the same product, you'll get the same result with less goodwill.
Once you know the root cause, there are five realistic options. Pick based on what the evidence says, not on how much the team wants a second launch day.
| What the post-mortem found | What to do |
|---|---|
| The product works, the message missed | Rewrite the messaging in customer language, test it with a small audience, then relaunch |
| It works for one segment and not the others | Narrow to the segment that's working and drop the rest for now |
| Buyers couldn't see why you're different | Reposition before you touch messaging or channels |
| The product wasn't ready for the people you sent to it | Fix the product first, then reopen the funnel in phases |
| No real demand from anyone | Stop, and go back to validation before spending more |
Narrowing is the option teams resist most, because it feels like shrinking. It's usually the fastest way back. A small group that buys and stays gives you proof, customer language and referrals. (I made the case for widening in stages in The Crescendo Product Launch.)
Stopping is the hardest call. Decision strategist Annie Duke, author of the book Quit, offers a simple test in her conversation with Lenny Rachitsky: knowing what you know now, would you start this today? If the answer is no, you're probably already late. Product leader Marc Abraham made that call on Settled Plus, a home-buying product at the UK startup Settled, and retired it within months once it was clear the people it depended on, including the legal professionals in the process, weren't ready to change how they worked.
Whatever you choose, set kill criteria for the next attempt before it starts: the number, at which stage, by which date, that tells you to widen, hold, or stop.
Explain it with the root cause, the evidence, and a recommendation, in that order, on one page. Leadership can handle a launch that missed. What erodes trust is a team that can't say why, or one that comes back with a new campaign before it has an answer.
What the page covers:
Own your part plainly. If you ran the launch, say which assumption you didn't test.
Ask your post-mortem questions before the next launch instead of after it. Most of what a post-mortem finds was knowable earlier, and the fix is to go looking for it on purpose before any budget goes out.
I wrote that side as its own piece: how to prevent a failed product launch. It covers the pre-mortem, the decisions to lock before you build a plan, and the go/no-go criteria that let you catch a problem days before it turns into a failure.
A product launch failed when it missed the success definition set before launch at one or more funnel stages, and that stage isn't improving after a full buying cycle. Check the numbers by segment first, because a launch that looks flat overall is often working for one group and failing for another.
Wait at least one full buying cycle. For a self-serve product, that's usually around 30 days of activation and repeat-use data. For a sales-led product, it's one average sales cycle. Leading indicators at each funnel stage will show you sooner which stage is off.
A slow launch is moving in the right direction at every funnel stage, just more slowly than planned. A failed launch has at least one stage sitting well below target and not improving. In retention, a slow launch levels off at some point, while a failed one keeps sliding toward zero.
Use the four After Action Review questions: what did we intend to accomplish, what actually happened, why did it happen that way, and what will we keep or change. Before the meeting, ask each stakeholder separately what they thought might go wrong before launch and whether they said it.
Include everyone who ran the launch or touches customers: product, marketing, sales, customer success, support, whoever owns the analytics, and a leader who can make decisions. Keep the room to about eight people and gather everyone else's input one-on-one beforehand.
Relaunch only if the root cause has changed. If the message missed, rewrite and test it first. If it worked for one segment, narrow to that segment. If the product wasn't ready, fix it and reopen in phases. If nobody showed real demand, stop and go back to validation.
Often, yes. Most failed launches are working for somebody, and narrowing to that segment gives you proof and customer language to relaunch from. Launches are hardest to save when there's no real demand from any segment, or when the team relaunches before finding the root cause.
Read more articles, explore services and pricing, or take the free Go-to-Market Readiness Assessment.