Weighted Word CountHow weighted word counts work
A weighted word count turns a CAT tool analysis into the number of words you are paid for at the full rate. This guide explains the bands, the usual percentages and exactly how this site maps Trados Studio and memoQ categories.
Why words are weighted
When a translation memory already contains a sentence, or a sentence repeats inside the project, the translator still has to read, check and sometimes adapt it, but the effort is lower than translating from scratch. Agencies and clients therefore pay a percentage of the full rate for each match band. The analysis report of your CAT tool counts the words in every band; a pricing grid assigns a percentage to each band.
The formula
Weighted words = sum over all bands of (words in band × % of full rate for that band).
Amount = weighted words × full rate per word.
This site rounds each band's amount to cents once per file (half up) and adds the rounded amounts, so every table adds up. With several files, each project band amount is the sum of the files' rounded band amounts, so the project total is exactly the sum of the per-file amounts.
A worked example
Full rate 0.10 EUR per word, default grid of this site:
| Band | Words | % of rate | Weighted | Amount |
|---|---|---|---|---|
| 101% / context match | 120 | 20% | 24 | 2.40 |
| 100% match | 340 | 30% | 102 | 10.20 |
| Repetitions | 255 | 30% | 76.5 | 7.65 |
| Fuzzy 95-99% | 410 | 60% | 246 | 24.60 |
| Fuzzy 85-94% | 230 | 70% | 161 | 16.10 |
| Fuzzy 75-84% | 160 | 80% | 128 | 12.80 |
| Fuzzy 50-74% | 95 | 100% | 95 | 9.50 |
| No match | 1,830 | 100% | 1,830 | 183.00 |
| Total | 3,440 | 2,662.5 | 266.25 |
Instead of 3,440 words at full rate (344.00 EUR), the job is billed as 2,662.5 weighted words (266.25 EUR).
Typical grids
There is no universal grid. Common agency grids pay 0–30% for context and 100% matches, 25–33% for repetitions, 50–70% for high fuzzy matches and the full rate below 75% or 85%. Many translators negotiate the full rate for 50–74% matches because those segments usually need retranslation. The default grid here (context 20%, 100% 30%, repetitions 30%, 95–99% 60%, 85–94% 70%, 75–84% 80%, 50–74% and no match 100%) is only a starting point: set the grid each client agreed with you.
How this site maps CAT tool categories
Both tools use their own names and sometimes their own ranges. The calculator normalises them into eight common bands:
| Common band | Trados Studio (XML) | memoQ (CSV) |
|---|---|---|
| 101% / context match | PerfectMatch (perfect), Context Match (inContextExact), Locked (locked) | X-translated, 101% |
| 100% match | 100% (exact) | 100% |
| Repetitions | Repetitions (repeated) + Cross-file repetitions (crossFileRepeated) | Repetitions |
| Fuzzy 95-99% | fuzzy ranges starting at 95-99% | 95%-99% |
| Fuzzy 85-94% | fuzzy ranges starting at 85-94% | 85%-94% |
| Fuzzy 75-84% | fuzzy ranges starting at 75-84% | 75%-84% |
| Fuzzy 50-74% | fuzzy ranges starting at 50-74% | 50%-74% |
| No match | New (new), fuzzy ranges starting below 50% | No match, Fragments |
Custom fuzzy ranges
If your Trados Studio analysis uses different ranges (for example 70–79%, 80–89%, 90–99%), each range goes into the common band that contains its lowest percentage: 70–79% counts as 50–74%, 80–89% as 75–84% and 90–99% as 85–94%. That way words are never billed as a better match than the tool guarantees. If you need the exact original ranges, use manual entry.
Internal fuzzy matches
Trados Studio can report internal fuzzy matches (matches against segments earlier in the same project). The calculator only counts them separately when the file's reported total includes them; otherwise they are already part of the new words and would be counted twice.
Locked segments
Locked segments are usually not translated again, so they are grouped with context matches. If your client pays them differently, use manual entry or adjust the context-match percentage.
Before you bill
Always compare the total words shown here with the total in your CAT tool. Exports can be customised, and a different layout may be read differently. When in doubt, enter the numbers by hand.