Validated
The pattern holds across enough languages, geographies and independent sources to be worth building against. This is where a property starts.
OSI
The tool we use to find the gap: a cultural tension that no character is answering yet, read across a dozen languages before anyone has given it a name.
By the time something is a trend, the space is taken
Most of the children's market moves after the fact. Something works, everyone can see it working, and the next eighteen months fill up with versions of it. The properties that end up mattering were started before that — when the need was already there but nobody had named it.
OSI reads what parents, teachers and psychologists actually say, in their own words, across a dozen languages, and looks for the same tension appearing in places that have no reason to be talking to each other. Almost everything gets discarded. What survives is where we start building.
The gap between a need appearing and the market answering it — that's the only place worth starting from.
Three states, and the discards are kept
Not every pattern becomes something to act on. What a pattern is classified as decides what we do with it — and the ones we throw out stay on the record with the reason attached.
Latent
Present, possibly significant, not yet consistent enough. It stays under observation and gets looked at again next cycle rather than being forced.
Discarded
Detected and rejected, with the reason written down. Keeping the discards is what makes the validated ones mean anything.
It isn't social listening
OSI doesn't measure popularity, track what is already trending, or read purchase data. Those tools describe what a market is doing now, which is useful for a lot of things and useless for this one: by the time a behavior is measurable at scale, the development window has closed.
It also doesn't generate ideas. It tells us where to look. What goes in that space is a creative decision, and it stays a creative decision.
Four times a year, and the archive is the point
OSI runs in quarterly cycles rather than as a one-off study, because a single reading tells you very little. The value is comparative: a signal that persists across four cycles is a different animal from one that spikes and fades, and you can only tell them apart by having both on file.
Q1January — March
Q2April — June
Q3July — September
Q4October — December
After several cycles the archive shows how a need emerged, consolidated or disappeared. That history is what tells us whether a window is opening or closing, which is a different question from whether a signal exists at all.
Because instinct was costing us properties
A couple of early properties were well advanced before it became clear that the need we thought we had found wasn't there. We stopped them. That's expensive in a way you don't notice at the time, and it's the reason this exists.
Every property we develop now starts from a validated signal. It's the first thing that happens on a commission, before anything gets drawn, and it's what keeps a commissioned property from coming out generic.