Powered by LanguageTool (via our server). Suggestions are automated—not a substitute for a human editor. Text is sent only for checking and is not stored by Trust Tool.
Unicode-aware word and sentence segmentation when your browser supports Intl.Segmenter; regex fallback otherwise.
Flesch-style scores run when the page locale is English (heuristic syllable model). Switch to English or use other metrics above for non-English copy.
First line length models a title tag; the meta preview uses the first 160 characters of your text flattened to one line (illustrative, not a crawler preview).
Green band ~50–62 characters (common SERP title range).
Green band ~150–160 characters (typical meta description window).
Sentence length mix: higher rhythm score means more variety between short and long sentences (entropy-based).
Percent = share of counted tokens. Toggle stop words for English to surface meaningful terms.
| Word | Count | Density % |
|---|---|---|
| No tokens to rank yet. | ||
When your browser supports Intl.Segmenter, we use locale-aware word and sentence boundaries. Otherwise a Unicode regex fallback splits tokens. Numbers and hyphenated words are handled as single tokens where the engine marks them word-like.
We bucket each sentence by word count (1 word, 2–6, 7–15, 16–25, 26–39, 40+). The score rewards a mix of lengths—similar to checking monotony—using a simple diversity measure on those buckets. It is editorial guidance, not a grade of quality.
Token rules, stop-word lists, and normalization vary. Here you can exclude common English stop words for the keyword table. Density is always relative to the tokens we count after that filter.
No. Analysis runs in your browser; nothing is sent to our servers for counting. Paste and clear stay on your device unless you use your browser’s own sync features.
No. Title and meta bands are rules-of-thumb (roughly 50–62 characters for titles and 150–160 for descriptions). Real SERP rendering depends on device, language, and query.
Your text is sent from the browser to our server, then to the LanguageTool engine (public or your self-hosted URL if configured). Results show grammar, spelling, and style hints with optional one-click fixes. It is not stored by Trust Tool; rate limits apply to protect the shared service.