Product launch metrics. What actually tells you a launch is working.

The short answer. Clearing up the failure-rate myth first.

Before product launch metrics are worth discussing, the number that scares most founders off tracking them properly needs correcting. The figure quoted almost everywhere, that 80, 90 or even 95 percent of new products fail, traces back to a Harvard Business School classroom reference from Clayton Christensen. That number bundles abandoned concepts and ideas that never left the whiteboard in with products that genuinely reached a market. Isolate only the ones that actually launched, and later analysis of the same ground puts the real failure rate closer to 35 to 45 percent, still a meaningful risk, just not the near-certainty the headline figure implies.

That distinction matters because it changes what product launch metrics are for. If failure were close to certain regardless of effort, tracking metrics would be closer to recording an outcome than influencing one. At 35 to 45 percent, the metrics become genuinely useful: an early warning system that lets a team adjust while a launch is still in motion, rather than a post-mortem written after the fact.

How it works in practice. Three windows, not one number on launch day.

Pre-launch: demand signals. Waitlist sign-up to conversion rate, qualified pipeline generated before launch day, or pre-orders against a forecast. These tell you, before a single unit ships or a single account goes live, whether demand is real or whether the launch is about to test a market that was never actually asked.

Launch week: adoption velocity. For a digital product, this is activation rate, the share of new sign-ups who actually reach a meaningful first action, not just registration. For a physical or hospitality launch, it is initial sell-through against the first batch or the first month's covers. Either way, this is the earliest real signal of demand, visible within days rather than the quarter it takes for a 90-day revenue figure to arrive.

Post-launch: revenue attainment. Actual revenue against the target set during planning, checked at 30, 60 and 90 days, alongside early churn or return rate and whatever satisfaction signal fits the category, NPS for software, repeat purchase for retail, rebooking for hospitality. This is the window that tells you whether early adoption translated into a sustainable result rather than a curious first wave. Choosing the right satisfaction signal matters more than it sounds: a software launch judged on repeat purchase has no equivalent event to measure, and a hospitality launch judged purely on NPS misses the much more commercially direct question of whether a guest actually rebooked. Match the signal to how the category genuinely earns repeat revenue, not to whichever metric is easiest to pull from an existing survey template.

PDMA's Comparative Performance Assessment Study, one of the more rigorous and long-running looks at new product performance, has surveyed hundreds of business units across dozens of countries over several editions. Its consistent finding is that firms running a disciplined, metric-tracked launch process outperform firms launching on an ad hoc basis by a wide margin, not because the product itself is better, but because problems in each of the three windows above get caught and corrected while there is still time to act on them.

The three windows, side by side. What to watch and when.

Decide who owns each window before launch day, not during it. Pre-launch demand usually sits with marketing or sales, launch-week activation with product or operations, and post-launch revenue with whoever holds the commercial target. Naming an owner for each window, rather than leaving the whole launch under one person's general watch, is what makes the three-window structure something a team actually uses day to day rather than a framework that lives only in the planning document.

A launch rarely fails at all three stages at once. Most commonly, one window looks healthy while another quietly does not, which is exactly why tracking a single blended number across the whole launch hides more than it reveals. A strong pre-launch demand signal followed by weak activation points at onboarding or positioning, not demand. Strong activation followed by fast early churn points at product fit or expectation mismatch, not acquisition. Knowing which window broke tells you what to fix; a single launch-day number never does.

What good looks like. A worked example from a Dubai market-entry launch.

I worked on a market-entry launch for a hospitality-tech product entering the UAE, where the team had, like most first launches, built a single success measure: bookings in the first quarter. That number alone gave no way to tell, in week one, whether the launch was working or heading for trouble. We built the three-window structure instead. Pre-launch, we tracked qualified demo requests against a target set from the pipeline the sales team had already built in the two months before launch. Launch week, we tracked activation, defined as a property actually completing setup and taking its first live booking through the system, not simply signing up for a trial.

The activation number came in well below where the pre-launch demand signal had suggested it should, inside the first ten days. That gap, caught in week two rather than discovered in the quarter-end numbers, pointed to an onboarding step that was taking properties far longer to complete than expected, not a demand problem at all. Fixing the onboarding flow rather than spending more on acquiring new sign-ups was the right call precisely because the metric that mattered was visible early enough to act on it. Waiting for the 90-day revenue figure to reveal the same problem would have cost most of a quarter of a wasted push on the wrong side of the funnel.

The decision rule I now use with every launch I work on is simple: if the leading metric for a window misses by more than a small margin, stop and diagnose before pushing harder on the activity that got you there. Spending more on demand generation when the real problem is activation just produces a bigger pile of sign-ups stuck at the same broken step, and it is tempting precisely because adding spend feels like progress while fixing an onboarding flow feels like admitting something went wrong. The teams that get this right treat a weak early signal as information to act on immediately, not as a result to explain away until the next review meeting.

Pitfalls to avoid. Where product launch metrics go wrong.

The first pitfall is tracking only vanity signals: press coverage, social mentions, sign-up counts with no activation attached. These feel like progress and photograph well in a launch recap, but none of them reliably predicts revenue or retention, and a launch can generate plenty of both while quietly failing on the numbers that actually determine whether it worked. A launch with strong coverage and weak activation has not failed quietly, it has failed loudly in exactly the place nobody was watching.

The second is waiting for the 90-day number before checking anything. By the time a weak quarter is visible in revenue, the window to fix onboarding, pricing or positioning cheaply has usually closed, and the fix becomes a relaunch rather than an adjustment.

A related version of the same mistake is choosing a launch date and treating the metrics plan as something to build afterwards, once the dashboard is needed. Decide the three or four numbers that matter, who owns watching each one, and what a weak signal in each will trigger, before the launch date is fixed, not after. A measurement plan built under the pressure of a launch already under way is rarely the same quality as one built with a few calm weeks of planning time behind it, and the gap shows up exactly when it matters most: in the first fortnight, when a weak activation number needs a decision within days, not a meeting scheduled for next month.

The third is setting one target for the whole launch rather than a target per window. A launch that hits its pre-launch demand number but misses activation badly is not failing the same way as one that activates well but churns fast after 60 days, and treating both as a single pass or fail score hides which part of the launch actually needs attention.

Track one or two metrics per window, decide in advance what a weak signal in each one means you will change, and review activation or sell-through inside the first fortnight rather than waiting for a quarterly number to tell you what was already visible weeks earlier. For the planning work that sits upstream of these metrics, our guide to a product launch strategy that works covers sequencing the launch itself, and the broader go-to-market motion guide covers how a launch fits into the wider commercial plan around it.