Security & architecture

How Qonva handles your uploaded data.

Current beta architecture · 24 September 2026

Raw CSV contents are analyzed in your browser.

When you select a CSV, Qonva reads it using browser-side JavaScript. The current analysis engine runs in the browser and does not send the raw uploaded file contents to Qonva application servers.

1. Browser-side processing

Conversation, CRM and optional marketing CSV files are parsed locally in the browser. The current beta analysis path does not use an external language-model API on your raw chats.

2. What Qonva stores

For real analyses, Qonva may store limited derived statistics such as signal category, sample size, close-rate comparisons, observed lift, confidence measurements, revenue-per-lead comparisons, recommendation state and measured outcomes. These records support your company-specific Sales Brain and product analytics.

3. What Qonva does not store from uploads

The learning tables are designed not to retain raw message text, raw uploaded files, lead IDs, rep names, campaign names or creative names.

4. Cross-company benchmarks

The Learning Network is opt-in. Benchmark results are aggregated and require a minimum participating cohort before they are displayed.

5. Authentication and database access

Supabase provides authentication and database services. Protected application data uses authenticated access and row-level security policies. Admin analytics require an additional admin authorization check.

6. Product analytics

PostHog is used for limited product analytics. Autocapture and session recording are disabled in the current implementation. Qonva configures analytics persistence in local browser storage rather than analytics cookies and does not intend to send raw uploaded customer conversation contents to PostHog.

7. Infrastructure

Vercel hosts the web application and may process standard infrastructure logs. Supabase stores account information and derived application data. These providers may process data outside Malaysia depending on their infrastructure and project configuration.

8. Limits of this architecture

Browser-side processing reduces server exposure to uploaded data, but it does not eliminate every security or privacy risk. Users remain responsible for access to their own devices, browsers and exported files.