AI models serve two types of users:

**Logged-In Users**

- These users sign in with an account (e.g., ChatGPT Plus, Google account). Logged-in sessions maintain persistent context, including stored preferences, search history, and account-specific instructions. Because the model can reference this persistent context, persona optimization in Gumshoe yields more substantial and consistent performance gains.

**Non-Logged-In Users** _(Guest Sessions)_

- These users interact without signing in. Guest sessions have no memory once the conversation ends. The model can only use temporary context, such as:
  - The text in the current session
  - Metadata like language or location
  - The system's default instructions

Once the session closes, all of this disappears. Persona optimization in these cases still matters, but mainly shapes tone and flow within a single interaction rather than building long-term influence.

## **Why does this matter for Gumshoe Personas?**

Gumshoe Personas simulate how AI models generate responses for various user types. But the effectiveness of persona-driven optimization depends on whether a model is dealing with a logged-in or non-logged-in user:

- **Logged-in Users:** Personas can significantly enhance performance because the model can connect answers to stored user preferences. These contexts often include enterprise users or frequent customers.
- **Non-Logged-in Users:** Persona shaping helps adjust how your brand appears in tone or structure during a single session, but won't persist. These contexts are more common for casual, exploratory users.

## **What's the ROI of persona optimization?**

Since no public data shows the exact split of logged-in versus guest usage, Gumshoe draws on academic sources to assume the following pattern:

- Casual and exploratory usage (e.g., consumers testing ChatGPT, Gemini, or Perplexity): often non-logged-in.
- Enterprise and frequent usage (e.g., teams with accounts, power users): usually logged in.

This means the highest ROI for persona optimization is in logged-in contexts, where the AI has persistent context to connect your brand with a persona's needs repeatedly.
