Word Count Cutter: plan what to cut when you're over the limit

Run the cut planner on your draft

Paste the draft below — the plan works best on 200 words or more, and up to 20,000 words is accepted. Set the word limit by typing it or tapping one of the preset chips, and the status line tells you how far over or under you are. Hit Build cut plan and the six category cards fill in: instance counts, estimated words saved, and an expandable list of every match with its own keep-or-cut toggle. The solver line underneath does the arithmetic and states the projected count in plain words. If you want to see the behaviour first, load the demo draft — an invented, illustrative essay that runs over a 500-word limit. Your text is never saved; only your limit and category choices are kept in this browser (key csw-wcc-v1), and Reset removes them.

One pass through the plan usually takes fifteen minutes on a 1,500-word draft. Work the categories in order of savings, but read each instance before cutting — the toggles exist because heuristics misfire, and the final count is done by you, not by this page.

Why drafts run over the limit

Drafts rarely go over the limit because every sentence is precious. They go over because academic prose grows padding the way a kitchen drawer grows utensils: one hedge here, a doubled pair there, a verb buried inside a three-word phrase, a clause that re-explains what the noun already said. None of it is ungrammatical. It is just expensive, and the bill arrives when the counter reads 1,640 against a limit of 1,500.

The padding clusters into six habits, and they are the same six whether the draft is a lab report, a case study or a personal statement. Each has a fixed, visible rule behind it — which is the only reason a page like this can count it honestly:

Here is the distinction the whole page rests on: wordiness is countable, importance is not. A machine can count how many words a phrase would shed, and it cannot tell you whether the sentence deserved to live. That is why this tool stops at the estimate — and why, while you edit, the word counter is the natural companion, keeping a live tally and reading time next to your revised text.

Hedging openers: the sentence that starts late

A hedging opener is a small speech at the door of a sentence. "It is important to note that sales fell" spends six words announcing the sentence before the claim arrives; "sales fell" does the job in two. These frames multiply under deadline pressure because they feel careful — and sometimes they are, when the uncertainty is real. The problem is density. A draft where every second sentence opens with a disclaimer reads as though the writer is apologising for having claims at all, and each apology is billed in words.

The scanner counts fifteen opener frames, and the list is deliberately about framing — the way a sentence opens — rather than about the hedge words scattered mid-sentence. (Style texture is a different lens; the AI writing tells checker reads hedging as a rhythm problem, not a word-count problem.) Cutting a frame needs one small mechanical care: whatever followed "that" becomes the opening of the sentence, so it gets a capital letter. Three before-and-after pairs, all invented and illustrative:

One caveat belongs in bold: keep the hedge when the uncertainty is the point. "The evidence suggests, though it cannot yet prove, that the effect fades by week six" is honest writing, and the plan is not asking you to sand it flat. Ration the frames rather than banning them — one deliberate hedge per genuinely uncertain claim, chosen on purpose.

Redundancy pairs and filler words

A redundancy pair is two words doing the work of one. "Past history" — history is past by definition. "End result" — a result is an end. These pairs survive because they sound emphatic, and emphasis is exactly what they fail to add. The scanner keeps a list of eighteen of them, each worth one saved word, and they are the cheapest cuts on the page: removing the second word of a pair has never damaged a sentence yet.

Filler words are the same disease with a single symptom. "Very unique", "really quite basic", "it was literally overnight" — the intensifier promises emphasis and delivers syllables. Sixteen common fillers are counted, one word each. Two cautions, both visible inside the tool: "just" sometimes means "only" and sometimes means "fair", as in "a just outcome", so judge each hit instead of mass-deleting; and "quite" strengthens in British usage ("quite brilliant") while it softens in American usage ("quite tired"), which is why it sits on the list without an automatic verdict. Four pairs, with the keeper:

One word per instance sounds trivial until the instances stack. Thirty fillers and a dozen pairs is roughly forty words — a third of a paragraph — bought back without touching a single idea, citation or transition.

Nominalizations: the long way to say a verb

A nominalization is a verb wearing a costume. "The committee made a decision" is "the committee decided" with two extra words and no extra meaning. "We conducted an analysis of the results" is "we analysed the results". Drafts lean on these constructions because they sound formal, but formality is not the same as force: the buried verb is usually the strongest word in the sentence, and digging it out pays twice — the phrase shrinks and the sentence wakes up. The scanner counts fifteen patterns, and each match displays its swap, so the fix is shown, not just the fault:

Wordy verb phrases: the long road to a short word

A close cousin is the wordy verb phrase: "is able to" for "can", "has the ability to" for "can", "due to the fact that" for "because". Twelve are counted, each shown with its one-word replacement. These are the safest cuts in the whole planner, because the replacement is a true synonym — the sentence survives word-for-word except for the words you keep. When the swaps remove a connective that was doing real work, replace it from a source with actual range, like the transition words sorted by job, rather than reaching for "moreover" on reflex.

The "that/which + be" clause

The last category is a structural trim: "the report that was released in May" is "the report released in May". When "that" or "which" is followed by "is", "are", "was" or "were" and then a participle, the two small words can usually go, and the scanner counts two saved words per instance. The rule is narrow on purpose — a participle must actually follow — because "the finding that is central to the argument" is doing different work than "the data that were collected in May". It will also miss irregular participles ("the manuscript which was put aside" slips past; "which was written" does not), so treat the list as a shortlist for your own ear rather than a verdict.

Where cutting hurts: voice, citations, evidence

Every category above is padding, which makes it tempting to conclude that cutting is always safe. It is not. Three things look cuttable and usually are not, and a plan that ignores them produces a shorter draft that is also worse.

• Evidence. A specific example, a number, a case — this is the load-bearing wall of an argument, and it is easy to mistake for decoration when you are skimming for words to kill. Cut the claim that has no evidence behind it, not the evidence that carries a claim.
• Citations. A quotation trimmed to a stub, a paraphrase stripped of its attribution, a reference list squeezed to save words — each of these turns a length problem into an integrity problem. If your rubric excludes references from the count, the honest fix is moving citations outside the counted body, not thinning them inside it.
• Voice. The sentence that sounds like you — the aside, the comparison only your course could produce — is the first thing any auto-trimmer deletes, because it is the least generic thing on the page. Protect it. This page will never generate a replacement, because there isn't one.

A practical order of operations: run the plan, take the mechanical savings, and only then consider structural cuts — merging two paragraphs that make the same point, retiring your weakest example, compressing a long quotation into a short one with a page number. Mechanics first keeps the scalpel away from the parts that make the draft yours. And once the survivors are in place, it is worth checking they still read cleanly — the readability checker scores exactly that.

The manual cut pass: six cuts that always work

The scanner finds the padding; this list is the part done by hand. Tap each line as you finish it — the meter keeps score:

• Replace every "in order to" with "to" — two words gone, every single time it appears.
• Delete the opener frames outright: it seems that, it is worth noting that, as a matter of fact.
• Keep the first word of each doubled pair: past history, end result, each and every, revert back.
• Turn three nominalizations into verbs: make a decision becomes decide; conduct an analysis of becomes analyse.
• Drop "that was" or "which is" before a participle: the report that was released becomes the report released.
• Re-count the draft, then read the opening paragraph aloud and cut whatever you stumble over.

What this planner cannot do

The honest limits, stated plainly, because a tool that plans your edits should have to earn that trust:

• It plans; it does not edit. No sentence is rewritten, replaced or generated. If the draft is to get shorter, your hands do it — the plan only tells you where the slack is.
• The lists are curated, not exhaustive. Seventy-seven patterns catch the common padding. They miss idiosyncratic wordiness, regional idioms and the specific tics of your own drafting, and they will flag some legitimate uses — "just" meaning fair, "quite" as a genuine intensifier. The instance list is where that judgement happens.
• The estimate is arithmetic, not prophecy. Savings assume clean rejoins. Sometimes a cut costs a word somewhere else — a needed connective, a capital, a re-spelled name. Re-count after editing, and expect small drift in either direction.
• It cannot judge importance. Counts rank phrases by category, never by meaning. A flagged sentence may be the best one in the draft; the keep/cut toggle on every instance exists for exactly that case.
• Under the limit is not automatically better. If the plan drops you far under, spend some of the credit on one more example or one more source rather than submitting thin. The limit is a ceiling, not a target score.

If you want more tools in the same private, in-browser spirit, the free writing tools hub collects them — counters, citation builders, the reverse outline — and the study guides hub holds the longer craft guides behind them. Cut early; padding is never cheaper to remove than the night before.

Quiz: five questions on cutting to the limit

Five questions on what the planner does, what the categories catch and where cutting hurts. A wrong pick reveals the correct one and locks the question — the tool above is fair game.

Frequently asked questions

Did the cut plan find enough words to get you under the limit? Yes Not really