How it works
The checker scans for three classes of low-information phrases: fillers ("basically", "actually", "in order to"), hedges ("might", "could", "generally") and weasel attributions ("experts say", "studies show"). Every match counts against your density score. The goal is an unhedged, confident, sourced voice - unambiguous assertions backed by named sources.
Fix patterns
- Weasel → named source - "experts say" → "according to [named researcher / institution]".
- Hedge → number - "often" → "in 68% of cases".
- Filler → cut - "It should be noted that GEO is basically…" → "GEO is…".
- Pair with the Fact Density Analyzer - raising facts and cutting fillers work together.
Why AI assistants punish hedging
Assistants are built to minimise hallucination risk. When they choose which content to quote, they actively look for unambiguous, sourced claims - because hedged content can be misinterpreted and cause hallucinations downstream. Unhedged, confident, specific writing is a cheat code. Full pattern in our content-to-cite guide.