Repetition, filler content and keyword density: understanding text analysis
Repetition, filler content and keyword density describe different properties of a text. They can help you find repeated words and edit your writing, but they are not an overall quality score. No percentage alone tells you whether an article meets the reader’s needs.
How the three metrics differ
- Word repetition index depends on how often the most frequent word occurs.
- Filler content is the proportion of words excluded from the core by statistical filtering rules.
- Word frequency, or density, is a word’s share of all words or of the core.
Start with text analysis, select the language and decide whether to group word forms using morphology. Use the same settings when comparing two drafts.
How Istio calculates the word repetition index
The formula is the square root of the highest occurrence count for a single word. Stop words are included. If the most frequent word appears 9 times, the score is 3. If it appears 8 times, the score is 2.82. Istio truncates the value to two decimal places rather than rounding it.
This is an absolute repetition score, not a percentage: the same 9 occurrences produce a score of 3 in both a short and a long text. An empty word list produces 0. Do not treat this score as a keyword-stuffing percentage or assume that another service uses the same formula for its repetition metric.
How filler content works
Istio’s core consists of the words left after removing stop words for the selected language and words shorter than three characters. For ordinary text with a non-empty core, filler content is calculated as: (all words − core words) / all words × 100%.
For example, if the filter excludes 40 out of 100 words, 60 core words remain and filler content is 40%. The final score discards the fractional part of the percentage.
There is a special case in the current implementation: if the core is completely empty, the interface may show 0. Do not interpret that result as “perfect text with no filler”; review the text itself and whether it contains meaningful words.
A stop word is not necessarily unnecessary. Prepositions, conjunctions and pronouns connect ideas. A short word can also carry important meaning. Removing every highlighted word just to lower the percentage often makes the text worse.
Word density in the text and in the core
For a single word, overall frequency is: number of occurrences / total word count × 100%. Three occurrences in a text of 40 words produce 7.5%. If the core contains 20 words, those same three occurrences represent 15% of the core.
The two columns use different denominators, so they should not be treated as one metric. Frequencies in Istio’s tables are truncated to one decimal place. Do not automatically apply the formula for a single word to the occurrence count of an entire phrase.
Why morphology changes the result
Morphology can group forms of a word. Without it, different spellings are counted separately; with it, their frequencies may be combined. For example, “text” and “texts” may belong to one group. The result depends on the language model and the particular words.
This affects the word list, the most frequent word, frequencies and the composition of the core. Text language, stop-word filters and word-form processing help explain why different services may return different numbers.
Edit your text using the analysis
- Read the text and identify the reader’s task it should address.
- Review frequent words and find repetitions that make reading harder.
- Inspect highlighted passages: remove empty phrases rather than necessary connecting words.
- Compare drafts using the same language and morphology setting.
- Finally, check spelling and reread the text.
Is there a correct percentage for search rankings?
Istio does not set a universal target for every genre. Acceptable repetition may differ between instructions, legal writing and product descriptions. Do not replace a precise term with an arbitrary synonym just to change a number.
Google describes excessive keyword stuffing as a violation of its search spam policies. That does not imply a universal “safe percentage”. Istio’s metrics help with editing; they do not guarantee search rankings.