How this works

Methodology & reliability

Workflow

  1. Input. Client name or topic, the date range to focus on, and an optional AI-context note.
  2. Discover. Review discourse across social platforms, news and the wider web - in all languages - to surface what happened around the client in that window.
  3. Prioritise. Distil the noise into the top 3 topics most likely to be interesting or risky for the client.
  4. Report. For each topic: a 2-3 line description, the platform(s) it was mainly seen on, why it matters for the client, 10 scan-ready keywords / hashtags, and supporting links.
  5. Scan. Feed those keywords into the scan builder to get platform-correct queries for broad monitoring in Cyabra's system.

Platform search rules

Each platform in Cyabra's system has its own search limits. The scan builder applies these automatically.

X (Twitter) AND / OR operators. Terms OR-grouped, phrases quoted, anchor AND-ed in.
Instagram Hashtags only. Each term becomes one hashtag (spaces / punctuation stripped).
TikTok Keywords only, matched as exact quoted phrases (one search per term).
Facebook Keywords only, matched as exact quoted phrases (one search per term).

Reliability principle

Findings must be based on data that was actually observed - never invented. Results may come from a sample of posts, comments and articles, but that sample should be a solid foundation that returns strong results when expanded into broader monitoring in Cyabra's system. Always attach the specific articles and posts a finding rests on so the client can verify it.

Scope of this tool

This app is a standalone analyst workbench. The scan builder is fully functional and runs entirely in your browser - no data leaves the page. The discovery report is a structured template you fill with your own investigation.

Automated live scanning of social platforms and news requires approved data connectors and credentials, which are intentionally not wired into a standalone demo app. That is the natural next step if you want the discovery step automated rather than analyst-driven.