Managing translation resources efficiently and optimizing the usage of processed words ensures cost-effectiveness.
As soon as a file is uploaded to Phrase TMS, the number of TMS processed words is calculated as follows:
A x B
= TMS processed words
Where:
-
A
is the total number of words in the source language in the uploaded content.-
If the source language is a character-based language (e.g. Japanese, Korean, Simplified or Traditional Chinese),
A
is the total number of uploaded characters divided by two.
-
-
B
is the total number of languages into which the source language is translated.
Processed words are only consumed when content is imported as a job in Phrase TMS. No other additional processed words are consumed in any other function (locking segments, analysis, using translation memory matches, etc.) nor do these functions reduce the number of processed words.
Example
A customer has a document with 5,000 source language words (A)
which are being translated into 5 languages (B)
. Therefore, the total amount of processed words equals:
5,000 words (A)
x 5 target languages (B)
= 25,000 processed words
Note
Use of Job Sync (adding languages) will impact total number of managed words.
Continuous jobs, projects, and update of source
If a source file is changed in continuous jobs, all words in new or altered source segments are counted as new and included in the TMS processed words count. Unchanged source segments within the same job are not included in the count, regardless of any updates to the target segments.
TMS processed words are deducted from the organization that imports the job.
If a buyer shares a project with a vendor from a different organization, the words count against the buyer's limits. The words only count towards the vendor's limits if they import additional jobs to the project. This applies to shared jobs as well.
A vendor will not see any words counted towards their TMS processed words limits when assigned to jobs shared by their buyers.
Best Practices for Word Management
To optimize the usage of processed words in TMS, the following strategies are recommended:
-
Regular audits to optimize content
Regularly review and eliminate unused or obsolete source content. Eliminate languages from project templates or any content repositories monitored by Automated Project Creation or other integrations.
-
Use continuous projects and jobs
Continuous projects and jobs provide significant savings compared to creating new projects and jobs each time.
-
Remove and avoid duplications
Ensure the content management system or processes do not inadvertently duplicate text across projects or departments. Be careful when importing overlapping sets of files.
-
Update source instead of reimport
If the original files uploaded to create a job have been slightly modified, select the job(s) and use Tools/Update source to update only the edited segments. This is also available via API.
-
Estimate Where Possible
When unsure if a large project will proceed, consider the following:
-
Avoid processing the documents and provide high-level cost estimates before uploading.
-
If translation memories are missing or offer minimal leverage, create jobs for one target language and use the respective analysis to estimate the rest.
-
Analyze a representative subset of files, extrapolate the results, and import the remaining files only after receiving the order.
-
-
Use Phrase Portal
Phrase Portal consumes only Machine Translation Units (MTUs) instead of TMS processed words, as no projects or jobs are created.
-
Use technical tips to save words
-
When importing files, upload a truncated version of the files first.
-
Instead of uploading multilingual XML, Excel, or CSV files with custom import settings to create jobs and populate a translation memory, it is advisable to pre-process legacy content externally. This approach will produce a multilingual Excel file or ordered document pairs (in the same format) suitable for alignment.
-
For XML files, use XML profiles instead of a trial and error. AI assistants can help generate complex XPaths to test in the of the XML profile in TMS.
-
To test regular expressions (regexp) on short samples, use regexr.com, regex101.com, or text editor plugins. For practical tutorials, visit rexegg.com.
-
For files processed with OCR, verify the contents and layout before import to prevent the need for reimporting corrected versions.
-