> ## 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. # Persona migration > What happened to personas that existed before the new persona data model — nothing was lost, everything was migrated. If you had personas before the current persona model launched, **nothing was lost** — every existing persona, created through Brand Analysis, Search, or Asset Evaluation, was automatically migrated into the new data model. No action was required, no persona was deleted, and none of your analyses, search projects, or asset evaluations lost their audience scoping. This page explains how that migration worked, so you know what to expect if you're looking at a persona that predates the new model. Looking for the current persona model itself? See the [Personas overview](/concepts/personas). ## What "migrated" means Legacy personas had been stored as a simple name string (plus, for Search personas, a description, core needs, and pain points) — there was no structured demographic or behavioural data behind them. When the new model launched, every one of those personas was carried over and back-filled into the current structure: * The original persona name became the persona's **archetype name**. * **Brand**, **category**, **country**, and **language** were inferred from the parent analysis. * For Search personas specifically, **Core Needs**, **Pain Points**, and **Description** were mapped onto the new **product/service need**, **pain points**, and **summary** fields. * Where a persona's name carried a clear gender or age signal (for example, an explicit "Male"/"Female" token, or an age range like "25-34"), that field was populated automatically; ambiguous cases were left blank rather than guessed. * Personas named "General population" (any casing) were excluded — they don't carry enough of an audience signal to be useful in the new model. * Because an analysis can target multiple countries, a Brand Analysis persona was fanned out into one migrated persona per country. A `Worldwide` scope was mapped to `United States`, except when the analysis already included `United States` explicitly — in that case the `Worldwide` entry was dropped rather than duplicated. * Asset Evaluation personas, which had no country or language on the legacy model, were migrated with `United States` / `English` as fixed defaults. * The migration is deduplicated and idempotent: re-running it never creates or overwrites a persona that already exists. Migrated personas are marked internally as system-migrated and slot into the same Library as newly created ones — there's no separate "legacy" view to check. Open your Library and they're there, alongside anything created or imported since. Fields the legacy model never captured (internal compass, behavioural map, resistance, connection details beyond what Search personas carried) are left empty on a migrated persona rather than invented. Enriching a migrated persona with that context is a manual edit — see [More options](/concepts/more-options) — same as for any other persona. ## What's next How personas are structured, created, and kept up to date. Import, edit, duplicate, archive, and delete personas.