Most of the time, you are not trying to understand everyone. You are trying to understand one group. The audience for this campaign. The segment you are deciding whether to serve. The people a policy is meant to reach.
Until now, when you polled personas in Cambium AI, you got your whole library at once. That works when you have a handful. But the moment you have created personas for a few different audiences, every question means working through all of them to get to the group you actually care about.
We have changed that. You can now label personas as you create them, group them by audience, and poll just that group whenever you want to poll them or talk to them.
In the video below, we create a group for one audience, label it, and poll only those personas.
Say you are planning a campaign for Gen Z renters. The group you have in mind is specific. People in their early twenties, no car, spending more than a third of their income on rent.
In Cambium AI, you create that audience as its own set of personas, drawn from verified US public data so the group reflects how that population actually looks. You give them a label. From then on, when you want to hear from Gen Z renters, you poll that group and nobody else.
You can ask them what matters most when choosing where to rent. How much of their income goes on housing. Whether they would sign another year at their current place. The answers come back from the group you defined, the same day, without sifting through every other persona you have created.
The point is not the label itself. It is what the label lets you do.
You can keep your audiences separate and work with one at a time, so your questions go to the right people, and the answers are not averaged across groups that have nothing in common. You can put the same question to two different groups and see where they split. And you can do all of it before you commit a penny of spend, rather than after a campaign is already running.
That is the difference between guessing what your target audience thinks and hearing it from a group built to represent them.
These personas are not invented in a workshop or guessed at by a model. Every group you create is grounded in verified US public data, and every persona traces back to a source. So when your target audience answers a poll, you are not reading an opinion the model made up. You are seeing what the data suggests about a real population.
That is the part that makes testing against them worth doing in the first place.
Set up your target audience as a group, label it, and pull just those personas the next time you have a question for them. If you already have personas in Cambium AI, you can start grouping them now.