Data
What Curious knows, and where it comes from.
Personalization is only as good as its context, and context is only as good as its governance. This page explains what data Curious uses, what it never uses, and how enrichment works on the Enterprise tier.
Your data
First-party context does most of the work.
Curious learns from what your product already produces: the events it emits (purchases, refunds, support tickets), the attributes you pass (plan, company size, industry), and the situation of each decision (device, page, locale). For most customers, that's all the context personalization needs.
Two rules govern how it's used. Identifiers like emails, names, and phone numbers are never used as model features; they exist for attribution, not prediction. And models only use information that existed at the moment of each decision, never anything that arrived after.
Your data trains your models. It is never pooled into other customers' models.
Enrichment
Enterprise adds licensed context.
Some decisions deserve more context than your product can observe on its own. On the Enterprise tier, Curious can enrich user context from licensed third-party sources: firmographic data for B2B surfaces (company size, industry, growth signals), property data for real-estate and home-services surfaces, and market-level geographic context. Pre-loaded CRM data and identity resolution connect what you already know about a customer to the decisions they see.
Enrichment is enabled per surface, deliberately, and only from sources licensed for exactly that use. A B2B paywall gets firmographic context. It doesn't get everything we could buy.
Governance
The rules it runs under.
Context is only worth having if you can stand behind how it was sourced and used. Four rules apply to every source and every customer.
Licensed, not scraped.
Every enrichment source is a licensed provider under contract, used within the scope of its license and your agreements. If a source can't be used compliantly for your use case, it isn't used.
Purpose-bound.
Enrichment data informs your optimization models and nothing else. It isn't resold, repurposed, or pooled into other customers' models.
No credit data. No precise location.
Some data carries its own regulatory regime and invites exactly the scrutiny personalization doesn't need. Credit information and precise location are the obvious examples. We don't use them.
Deletable.
A user's identifiers, history, and enriched context live together. Delete the user and everything Curious holds about them goes too.
Context with a chain of custody.
Ask us exactly what we'd use for your audience, and what we wouldn't.