Draft
When you have read a draft many times, its wording can become invisible. Counting a few features can bring your attention back to the page. The aim is to notice choices, not to make every document follow the same numerical profile.
Start with repeated openings
Read the first few words of each sentence in a paragraph. “The study found…” can be helpful once. Used five times, it may make the paragraph feel like a list of disconnected reports.
Before: “The study found that the seedlings grew faster. The study found that light mattered. The study found that watering frequency had little effect.”
After: “Seedlings grew faster under stronger light. Watering frequency, however, had little effect in this experiment.”
The second version is shorter because it combines information. It is not automatically better: if the original sentences described different experiments, combining them could erase an important distinction. Check the evidence before revising the structure.
Sentence rhythm is more than variety
DraftLens counts the words in each sentence and calculates the coefficient of variation: the standard deviation of those counts divided by their mean. A larger value means sentence lengths differ more. It does not mean the prose is better.
Instructions may benefit from short, similarly sized sentences. A technical explanation may need one long sentence to preserve a condition and its consequence. If the rhythm feels monotonous, try changing a sentence for a reason: emphasize a conclusion, separate two steps, or clarify a relationship.
Sentence segmentation also has limits. Abbreviations, headings, quotations, and unusual punctuation can affect where software sees a boundary. Inspect the text rather than assuming the count is exact in every genre.
Repeated phrases can be purposeful
The analyzer looks for repeated five-word sequences and repeated paragraphs. Those patterns can identify accidental duplication. They can also find legitimate quotations, refrains, definitions, and required terminology.
Suppose a methods section repeatedly uses “the temperature of the solution.” Replacing each instance with a different phrase might make the writing less precise. Technical consistency is often more valuable than cosmetic variety. Edit repetition when it adds friction or no new meaning, not merely because it appears in a report.
Vocabulary ratios are length-sensitive
Lexical diversity here means unique normalized words divided by total words. A ten-word sentence can easily have a higher ratio than a two-thousand-word essay. That does not make the sentence more sophisticated.
Compare passages of similar length and purpose if you use this ratio at all. Proper names, specialist terminology, dialogue, and repeated function words influence it. Treat it as a description of the sample, not a target to maximize.
What readability formulas leave out
Flesch Reading Ease and Flesch–Kincaid Grade use sentence length and estimated syllables per word. They do not evaluate whether a claim is true, an argument is coherent, or a reader understands the subject. A short unexplained acronym may lower the estimated difficulty while making a paragraph less useful.
Our implementation estimates syllables locally. Names and uncommon words can be miscounted. Readability is withheld for short or clearly unsuitable input, but even a displayed number is approximate. For background on readability as one part of broader communication, see the CDC Clear Communication Index.
Finish with a reader’s question
Choose one paragraph and ask: What should my reader understand after this? Then inspect the tool’s observations against that goal. Remove accidental duplication, preserve needed terminology, and read the revised paragraph aloud. That process turns a set of numbers into an editorial decision.
Use the word counter for basic metrics and the writing analyzer for exact repeated spans. Neither tool grades your intelligence or identifies your author.
Put it into practice.
Take a fresh look at your draft with the related local tool.
Open writing patterns