79,573 products and the 12 breakouts that changed everything in Q2
Twelve products broke from the pack this quarter. Line up their signal-score curves against the other 79,561 and the same three patterns show up every time.
Twelve is the number that matters this quarter, but it only means something set against the other one: 79,573. That's the full corpus Launch Sentinel had scored through Q2, every product checked against its own history for the kind of divergence that gets flagged, not against some universal launch bar. Twelve cleared it. The other 79,561 didn't.
We pulled the twelve into one view and asked the obvious question: what do they actually share? Not the category, not the funding round, and not which outlet broke the story first. The pattern sits one level down, in the shape of the signal-score curve itself, and once you've seen it in one breakout, you start seeing it in all twelve.
What counts as a breakout here
Launch Sentinel doesn't grade products on an absolute scale. It grades them against their own trend line. A product sitting at a flat, unremarkable signal score for months that suddenly clears its own baseline by a wide margin, and keeps clearing it, gets flagged as a breakout. A product that was already strong and got a little stronger doesn't qualify. Neither does a product that spikes for a week and settles back to where it started.
That's a stricter bar than it sounds. Across 18 categories and 79,573 tracked products, twelve cleared it this quarter. Zoom out and the corpus underneath is anything but still: 353 products died in the same stretch, and 812 came back from the dead. More resurrections than deaths, in one quarter, which tells you more about how fluid this market actually is than any single top-line count can.
The trajectory that repeats
Line up the twelve against the field and the shapes diverge in one specific place. For the first several weeks of the quarter, the breakout cohort's median signal score tracks the rest of the corpus almost exactly: flat, unremarkable, nothing to flag. Then, somewhere around week five or six, the breakout line breaks away. It doesn't spike and fade. It holds.
That hold is the whole signal.
Illustrative aggregate pattern: the median weekly signal score of the 12 flagged breakouts vs. the field median across Q2 2026. Built to show the shape of the divergence, not per-product data.
What the twelve actually share
Three things showed up across all twelve, in some form. None of them is the thing you'd guess first.
The press spike comes first, but it isn't enough on its own
Coverage volume moves before signal score does in every one of the twelve. But plenty of other products spiked in coverage this quarter too, and faded inside two weeks. Those don't make the list. A spike is cheap. A held score isn't.
Category tells you almost nothing
This quarter's twelve span categories that hadn't produced a breakout in over a year, sitting next to categories that produce one most quarters. Bet on last quarter's winning vertical and you'd have missed most of these twelve.
The plateau outlasts the launch
One of the twelve is a workflow-automation tool that sat at the same modest signal score for over a year before anything moved. It picked up its first trade-press mention six weeks ago and hasn't given the new score back since. That's the shape that separates a breakout from a good week.
A press spike tells you a product got noticed. A held signal score tells you it got used.
Reading it against the corpus
Twelve out of 79,573 works out to roughly one in 6,600. Call that the base rate for a breakout in a single quarter, across 18 categories. It's a small number on purpose. Loosen the anomaly threshold and you'd catch more press spikes and call them breakouts; you'd also be wrong about most of them within a month.
The corpus context matters as much as the twelve themselves. The same quarter that produced twelve breakouts also produced 353 deaths and 812 return-from-the-dead transitions, evidence that most of the movement in a market this size happens without anyone noticing. The twelve are the loud exceptions. Everything else is baseline you only see if you're scoring the whole corpus, every week, against itself.
See your own market like this
Launch Sentinel scores 79,573 products across 18 categories every week and flags the ones breaking from their own history before the press cycle catches up.