What is a synthetic audience?

Audience without a profile

A conventional audience is a list of identifiers, assembled from things people did: pages visited, apps used, purchases made. It persists, it is stored, and it requires recognising the same person over time.

A synthetic audience holds none of that. It is assembled at request time from what the user has said in this conversation — the goal they stated, the constraints they gave, the stage they appear to be at. The definition is a description of a situation rather than a list of people.

The practical consequence is that a user can match a valuable audience with no history at all, and can stop matching it a minute later when the conversation moves on.

What it fixes

• No identifier dependency. Nothing to deprecate, because nothing persistent is used.
• No cold start. A first-time visitor is as targetable as a returning one.
• Current rather than historical. Someone who bought a mattress last year is not still in-market; someone asking about mattresses now is.
• Far less data retention, which changes the privacy posture materially — see privacy thresholds.

Where it fails

Three limits are worth being explicit about, because the concept is easy to oversell.

It cannot see outside the conversation. Loyalty, prior purchases, lifetime value — all invisible unless stated. For businesses whose targeting edge is their customer history, this is a real loss.

Interpretation is fallible. The model is inferring intent from language, and it will sometimes read enthusiasm as intent or research as buying. Conventional audiences fail differently, not less.

It is ephemeral by design. You cannot build a segment and reuse it next quarter, which breaks the standard audience-planning workflow. Planning has to be expressed as conditions rather than as lists.

How to plan against it

The translation from an audience brief is usually straightforward once the habit shifts: describe the situation rather than the person. "Small business owners aged 30–50" becomes "someone choosing a business bank account for a company with under ten employees." The second is both more precise and directly matchable against a conversation.

It also pairs naturally with intent density — the situation defines relevance, the density defines what it is worth.