Public persona library vs custom personas — which do I use?
The short answer: for most studies, use the public library — it's pre-built, covers common demographics, and is ready to use in seconds. Build custom personas when you specifically need to simulate your real customers, a niche audience the library doesn't cover, or a private group whose data shouldn't go near a shared library.
The two persona sources at a glance
| Public library | Custom personas | |
|---|---|---|
| Count | ~5,000 pre-built | As many as you create |
| Source | Built from anonymized research, user profiles, prior Scout runs | Your own data (interviews, CRM, Scout) |
| Quality | Stable, validated, fits general research | Quality depends on your data depth |
| Privacy | Shared across all users | Private to you (or your team) |
| When to use | 80%+ of studies | Specific gaps the library doesn't cover |
| Cost | Included in all plans | Included in all plans; Scout costs quota |
Most studies can use the public library exclusively. Reach for custom only when you have a specific reason.
When to use the public library
The library fits when:
- Your audience is a common demographic (urban professionals, students, young parents, fitness enthusiasts, etc.)
- The study is exploratory — testing a concept, sizing a market, generating hypotheses
- You don't have existing customer data and don't need to simulate specific people
- You want a fast turnaround — pick personas in 2 minutes, start a study immediately
It doesn't fit when:
- You're trying to simulate your actual VIP customers (no library has them)
- Your audience is a niche community the library can't represent (caving enthusiasts, ham radio operators, etc.)
- You have private customer data and need a Panel that reflects it
When to use custom personas
Three concrete scenarios where custom is the right call:
1. You have real customer interviews. Upload 5–10 transcripts. AI extracts personas with backgrounds matching your real customers. Now you can test new products, copy, or pricing against your actual user base — without re-interviewing them.
2. You're testing in a niche the library can't represent. Run Scout to observe public discussions about the niche (e.g., specialty coffee roasters, ultralight backpackers). AI builds 5–10 personas from what it sees.
3. You need a long-term Panel of specific people. Build a Panel of 10 personas that represent your ideal customer profile. Use the same Panel for every quarterly study. Over time, this becomes a reusable research asset.
Custom personas are private by default. They appear in your "My Library" tab, not in the public library search. Team plans can share them across team members.
How custom personas are built
Three ways:
Upload interview transcripts. Best when you have detailed 1-on-1 interviews. AI extracts background, attitudes, behaviors. One transcript = one persona (or merged across multiple short interviews).
Upload survey responses or CRM exports. Good for breadth. AI finds patterns across hundreds of responses and builds representative personas.
Run a Scout task. Best when you don't have your own data. Scout observes public social media and builds personas from observed behavior.
Each approach takes 10 minutes to 2 days depending on method. The output is added to your "My Library."
The cost-benefit picture
| Approach | Build cost | Reuse value |
|---|---|---|
| Public library | Free | Reusable across studies |
| Custom from your data | One-time upload | High if reused across multiple studies |
| Custom from Scout | Scout quota (small) | High if the niche is recurring |
The math: custom personas pay off when you reuse them across 3+ studies. If it's a one-off question, the public library is more efficient.
A practical rule of thumb
Default: public library. Pick 5–10 personas that span concerns. Run the study. Iterate.
Upgrade to custom when:
- The library doesn't have your niche
- You have specific people you need to simulate (VIP customers, target accounts, user research participants)
- You're running repeated studies on the same audience and want consistency
Mixing public and custom
You can mix in a single study. Example: 7 library personas + 3 custom personas built from your actual user interviews. The custom personas reflect your real audience; the library personas fill in adjacent segments you might expand into.
Common mistakes
Building custom when library would do. Custom personas are powerful but cost time. If the library has a good-enough fit, use it.
Building custom from poor data. Garbage in, garbage out. If your interview transcripts are short, formulaic, or outdated, the resulting personas will be too.
Treating library personas as "real users." They're AI approximations grounded in real data. Use them for direction, not for statistical certainty.
How private are custom personas?
- Private to you by default
- Not searchable by other users
- Not used to train any shared models
- Team plan: you can selectively share with team members
If you're working with sensitive customer data, custom personas + private workspace is the right combination.
Related: Persona Library, Scout, Panel, Interview, Discussion