The short answer
A synthetic tribe is a model of a real consumer tribe, built from documented observations about it. You can ask it questions, put a campaign or an offer in front of it and see where it agrees and where it splits.
It is not a chatbot playing a character, and it is not a survey. Every observation behind it carries a source, a population and a year, and it answers as several perspectives rather than one average voice.
What it is built from
Each tribe starts with research: published studies, public reporting and what organisations say about themselves. Every claim is archived with its original passage and sorted into one dimension of The Tribal Map: values and meaning, artefacts, spaces, influencers, stories, language, experiences and tensions.
The observations are also tagged with the perspectives they belong to. A tribe usually has a shared core and four to six perspectives that see things differently.
How it answers
When you ask a synthetic tribe something, each perspective answers from its own observations. The answer says where they agree, where they pull apart and what would change their view. A second model then checks the answer against the observations before you see it.
That makes it useful for the questions a single persona hides: who will object, who will wait, and why.
What it is not
A synthetic tribe is a simulation. Its answers are not statements by real people and not a forecast of what everyone in a tribe will do. It uses no personal data, and its voices are always labelled synthetic.
How well it holds up is tested blind against real outcomes it has not seen. Each tribe shows its result once that test is done.