What is a weighted word count?
A weighted word count (also called a CAT count, weighted count, or analysis-weighted volume) is a single billable word total calculated from the match-type breakdown produced by a Computer-Assisted Translation tool such as SDL Trados, memoQ, Wordfast, or Logoport. Instead of charging for every raw word in a document, repetitions and high-similarity matches are billed at a reduced rate — because the translator’s effort on those segments is lower.
Why translation agencies use it
CAT tools split a source document into segments and check each segment against a translation memory. A segment that already exists in memory (a 100% match) needs review, not re-translation. A fuzzy match at 85% needs partial editing. A no-match segment is full work. Charging the same rate for all three would either over-bill the client or under-pay the translator. A weighted word count fixes that by multiplying each tier’s word count by a percentage — the “weight” — then summing the result. The output is one number both sides can agree on.
The formula
A worked example
A 6,680-word Trados log breaks down as 1,240 repetitions, 860 100% matches, 410 in the 95–99% tier, 320 at 85–94%, 280 at 75–84%, 150 at 50–74%, and 3,420 no-match words. Applying the industry-standard weights (30% / 30% / 50% / 60% / 80% / 100% / 100%) gives a weighted count of 4,566 words — a 32% reduction from the raw count. At €0.09 per word, the invoice lands at €410.94 instead of €601.20.
Standard match tiers
Tier names and percentage weights vary slightly between agencies and CAT tools. CATCount lets you save your preferred scheme per client.
Common questions
Is a weighted word count the same as a Trados word count?
A “Trados word count” is the raw match-type breakdown produced by SDL Trados. A weighted word count is what you get after you apply percentage weights to each tier and sum the result. CATCount does that math for you.
Who decides the weights — the translator or the agency?
Both. The weights are part of the rate negotiation. Once agreed, save the scheme in CATCount and reuse it for every job from that client.
Does CATCount work for character-based languages (Chinese, Japanese)?
Yes — CATCount works at the match-tier level, not the character level. As long as your CAT tool produces an analysis log, CATCount can weigh it.
For everything before the CAT tool — counting words, characters, lines, and pages in the source document itself — pair CATCount with our sister product AnyCount. AnyCount is a dedicated word count tool that supports over 100 file formats out of the box: DOC, DOCX, RTF, ODT, PDF (text and scanned), XLS / XLSX, PPT / PPTX, HTML, XML, MIF, INX, IDML, INDD, FrameMaker, Visio, AutoCAD DXF, EPUB, MOBI, JPG / TIFF / BMP (via OCR), CSV, SRT, VTT and dozens more — including ZIP archives counted recursively. It produces accurate quotes in seconds, exports to Word / Excel / HTML / CSV, and handles batch folders, embedded text, footnotes, headers, comments, and text inside images. Together, AnyCount tells you how many words are in the source, and CATCount tells you how many of those words to invoice after CAT leverage — the two-step quote pipeline used by translation agencies worldwide.
Is the weighted count fair to the translator?
When the weights reflect real effort, yes. The repetition rate of 25–30% exists because reviewing a previously translated segment genuinely takes about a third of the time of translating it from scratch. The point of a weighted count is honesty in both directions — not a discount.