> ## Documentation Index
> Fetch the complete documentation index at: https://docs.shareofmodel.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Overview
> Narrowing an analysis to a specific audience — how personas are structured, created, and kept up to date.
A **persona** narrows an analysis to a specific audience. Instead of asking a model what it thinks about a brand in general, a persona-scoped prompt asks what it thinks *for someone in particular* — a value-driven parent, a luxury beauty shopper, an experienced developer. Up to 6 personas can be attached to a single analysis; each one adds its own audience-scoped variant of every prompt.
Personas are shared across modules — the same persona created in one module can be reused wherever personas are supported:
* **Brand Analysis** — personas narrow the perception and awareness prompts sent to LLMs.
* **Search** (AI Visibility) — personas narrow search queries and carry funnel-stage/intent metadata alongside them.
* **Asset Evaluation** — personas provide the audience lens creative assets are scored against.
## The persona model
Every persona is structured around the same set of fields, grouped into a few blocks:
| Block | Captures | Required |
| -------------------- | --------------------------------------------------------------------------- | -------- |
| **Profile** | Name, brand, category, country and language. | Yes |
| **Demographics** | Age range, gender, education, occupation, income, living context. | No |
| **Internal compass** | Job to be done, values and personality traits, motivations, strategic goal. | No |
| **Behavioural map** | Lifestyle and primary tasks, media consumption, a typical quote. | No |
| **Resistance** | Pain points, purchase barriers, core questions, key buying triggers. | No |
| **Connection** | Product need, why the persona loves the category, a summary. | No |
Not every field needs to be populated for every persona.
Even if we encourage well-defined persona, a persona created quickly from a short brief can have a thinner profile than one built from a full research pass, and that's expected.
This same profile also powers **Activation** — see below.
## Activation
Beyond scoping analyses, every persona automatically maps to the Google product categories that best reflect its profile — the same interest segments used for ad targeting (for example, *Beauty Products & Services/Makeup & Cosmetics* for a beauty-focused persona). Each matched category comes with the ad surfaces it can be activated on: Display, Gmail, Discover, Search, Shopping, and Video.
The mapping is derived automatically from the persona's existing fields — brand, category, and its demographic and behavioural signals — there's nothing extra to configure. You can see the full breakdown, grouped by product category, on a persona's **Activation** tab.
## Using personas in a creation flow
Every Brand analysis, Search project, and Asset Evaluation includes a dedicated Personas step. Rather than listing your whole Library, that step filters it down to a matching set for the analysis you're configuring — and lets you create a new persona on the spot if nothing fits.
The matching scope differs by module:
Personas are matched on **Brand + Country + Language**. Category is intentionally excluded from the match, so a brand's personas surface regardless of which category the analysis targets.
Same scope as Brand analysis: **Brand + Country + Language**. In Fast Mode, up to 3 matching personas are auto-selected and the Company Insights step is skipped; if none match, 3 personas are generated automatically (from Brand, Country, Language and Category) and saved to the Library.
Personas are matched on **Brand only** — Category, Country and Language aren't part of the match, since evaluations aren't scoped by locale.
If nothing matches, you can create a persona manually or generate one with AI without leaving the flow — it's saved to your Library and reused the same way in later analyses.
## Best practices
Right after setting up your workspace, create or import your qualified personas. It's easier to do this once, up front, than to build personas as you go while creating projects and analyses — even though that path remains possible.
Demographics like age, country, or language help scope a persona, but what makes it useful is the "why" behind the audience. A persona without a clear need or pain point behind it gives weaker signal.
A persona that tries to represent everyone ends up biasing nothing. One clear need per persona tends to give sharper signal than a broad, catch-all profile.
One well-qualified persona outperforms ten shallow ones — better signal, and more reliable across analyses.
AI-generated personas work well as a starting point, but they will rarely match a persona your brand has already researched and refined over time. If you do generate a persona with AI, take the time to review and validate the full profile before using it.
## What's next
Import, edit, duplicate, archive, and delete personas.
Have a live, in-character conversation with any persona in your Library.
What happened to personas that existed before the new data model.
How personas fit into a Brand analysis or Search project's scope.