Translation Memory explained: Save time and money in localization
Learn how Translation Memory works and why it is essential for efficient app localization. Reduce costs and ensure consistency in updates.
Advanced TM concepts
TM-Analytics
Track your TM performance:
Match rate in percent
Cost savings per project
Consistency scores
Context matching
Modern TM systems consider context:
Context metadata helps select the right translation.
"Save" (Button) → "Save"
"Save" (discount) → "Save"
TM maintenance
Over time, TM may require cleanup:
Regular maintenance keeps TM effective.
Remove outdated translations
Update terminology
Merge duplicates
Maximizing TM effectiveness
Best Practice #1: Consistent source text
Bad:
['"Click Save to continue"', '"Press Save to continue"', '"Tap Save to continue"']
Three different strings = three translations.
Good:
['"Click Save to continue"', '"Click Save to continue"', '"Click Save to continue"']
One string = one translation, reused three times.
Best Practice #2: Modular Strings
Bad:
"Welcome back, '{'{name}'}'! You have 5 new messages."
This entire string must match exactly.
Good:
['"Welcome back, \'{'{name}'}\'!"', '"You have \'{{count}}\' new messages."']
Each part can match independently.
Best Practice #3: Do not include context in keys
Bad:
['"home_screen_save_button": "Save"', '"settings_screen_save_button": "Save"']
Same text, different keys = no TM advantage.
Good:
"save_button": "Save"
One key reused everywhere.
Build your Translation Memory
TM across projects
Shared TM means that common phrases only need to be translated once:
"Settings" → "Settings" (translated in App A)
App B automatically uses the same translation
Same company, multiple apps?
Import existing translations
If you already have translations:
Upload your translated files
The system links source and target texts.
TM is populated immediately
Future translations benefit immediately
Start fresh
If you're new to localization:
Translate your first version as usual
TM is built automatically
Future updates reuse existing translations
Getting started with TM
If you have existing translations
Upload source files (English)
Upload translated files (German, etc.)
System builds TM automatically
Start benefiting immediately
If you're new
Register at shipglobal.dev
Upload your strings
Translate once
TM builds up automatically
Future updates are cheaper and faster
How Translation Memory works
The basic process
You translate: "Welcome to our app" → "Welcome to our app"
TM stores it: Both versions are saved as a pair.
Next time: When "Welcome to our app" appears again, TM suggests the German translation.
You save: No need to translate the same text twice (or pay for it)
Translation Memory in Practice
Example: App update workflow
150 changed strings (fuzzy match) → review the proposed translation
50 new strings → translate
Result: Instead of translating 5,000 strings, you translate only 50.
Your app v2.0 has:
4,800 unchanged strings (100% match) → free
How shipglobal.dev TM uses it
You only pay for truly new content.
Our system automatically:
Hashes every string for fast search
Looks for exact matches (100% free reuse)
Identifies fuzzy matches for review
Applies translations immediately
Match types
100% Match
Exactly the same text
"Save" = "Save"Fuzzy Match (75-99%)
Similar text
"Save changes" ≈ "Save all changes"No Match (0-74%)
New content
"Welcome aboard!"ROI Calculator
Ohne
€10 per month translation cost
1,000 new strings/month
€120 per year
Mit
€2 per month translation cost
100 new strings/month
€24 per year
Annual savings: €96 (80%)
TM vs Machine Translation
Best approach: Use both. TM handles existing content; MT handles new content.
Cost
Cost per word
Free for matches
Quality
Good, needs review
Guaranteed (it's your text)
Source
AI model
Your previous translations
Speed
Seconds
Instant
What is Translation Memory?
Translation Memory (TM) is a database that stores previously translated text segments. When you translate new content, the system checks for matches and reuses existing translations.
Think of it as autocomplete for translations—only smarter.
Why Translation Memory matters
2. Consistency
"Log out" → "Sign out" (Screen 1)
"Log out" → "Sign out" (Screen 2)
Without TM, the same phrase could be translated differently:
TM ensures consistent terminology across your entire app.
1. Cost savings
Ohne
Total: 15,700 paid translations
Version 1.0: 5,000 strings → pay for 5,000 translations
Version 1.1: 5,200 strings → pay for 5,200 translations
Version 1.2: 5,500 strings → Pay for 5,500 translations
Mit
Total: 5,500 paid translations
Version 1.0: 5,000 strings → Pay for 5,000 translations
Version 1.1: 5,200 strings → Pay for 200 new translations
Version 1.2: 5,500 strings → Pay for 300 new translations
Savings: 65%
3. Speed
Ohne
5,000 strings from scratch: hours
Mit
Translating 200 new strings with TM: minutes