That distinction has largely gone unchallenged.
Because historically, content hasn’t been easy to structure.
Especially video.
Video has always been treated as something to store, stream, or transcribe. Not something to systematically break down and use as a signal.
That’s starting to change.
We’ve been analyzing video at scale and structuring it into a set of signals that can be used the same way traditional data is used.
Things like:
- topic
- sentiment
- entities
- tone
- scene-level context
Not inferred from users.
Derived directly from what’s actually being watched.
What’s interesting isn’t just the data itself.
It’s what it reveals.
Across large-scale analysis, we’ve consistently seen that:
- content signals explain behavior in ways traditional data does not
- two audiences that look identical on paper behave differently based on content exposure
- brands and products appear in video far more often than they are captured in metadata
This creates a gap in most data stacks.
On one side, you have detailed user-level understanding.
On the other, you have an incomplete view of the environments influencing those users.
Most organizations try to bridge that gap with proxies.
Categories. Keywords. Contextual tags.
But those are approximations.
They don’t capture:
- tone
- narrative
- nuance
And that’s where the opportunity is.
Video represents one of the largest and least structured sources of information in the ecosystem.
Turning it into a usable signal layer fundamentally changes what can be understood, modeled, and activated.
The question is no longer whether video can be analyzed.
It’s whether organizations are incorporating it into their data strategy.
Because the ones that do will have access to something others don’t.
A direct line between content and behavior.
If you’re evaluating new data sources or thinking about gaps in your current stack, it’s worth looking at what video can actually contribute when it’s structured correctly.
