Quick Facts
| Topic | Detail |
| Full Name | Context Match (also: 101% Match, ICE Match) |
| ICE Stands For | In-Context Exact Match |
| Used In | CAT tools (Trados, memoQ, Phrase, Smartcat, etc.) |
| Match Score | 101% or 102% |
| What It Checks | Text + surrounding segments + document structure |
| Who Benefits | Translators, project managers, localization teams |
| Cost Impact | Often billed at 0–10% of normal word rate |
| Key Advantage | Highest reliability of any match type |
| Common Confusion | People think 100% is already perfect — it isn’t |
So What Is a Context Match, Anyway?
Let me start with something a bit mind-bending.
Imagine your CAT tool (that’s the software translators use) shows you a score of 101%. Your first thought? “Wait — how can anything score higher than 100%?”
You are not alone in feeling confused by that.
That number actually makes perfect sense once you understand what it means. A context match isn’t just a perfect word match.It is a perfect word match plus evidence that the sentence is positioned exactly where it has always been, with the same text in front of it, in the same document structure, and surrounded by the same neighbors.
Think of it like this. Say you find your house key on the floor. That’s a 100% match — right key, right shape, right teeth. But a context match would mean it’s also in the right drawer, in the right room, in your own house. Everything checks out. Every single thing.
That extra layer of certainty? That’s what makes it go beyond 100%.
See also “GTA 5 Cheat Codes: The Complete Guide Every Player Needs“
How Translation Memory Works First
Before context match makes sense, you need to know about Translation Memory. Let’s call it TM for short.
A TM is basically a giant notebook. Every single sentence a translator ever confirms gets saved in it — the original text and the translation. When a new document comes in, the software opens that notebook and starts scanning.
Did we translate this sentence before? Yes? Pull it out and show it to the translator.
That saved sentence is called a segment. Segments are the building blocks of everything in a CAT tool. The whole document gets broken into these little pieces, one at a time.
Now here’s where it gets interesting. When the tool finds a segment that is 100% identical to something already in the TM, it shows it as a perfect match. But — and this is the key thing people often miss — a 100% match only checks the words. It doesn’t check what comes before or after that sentence.
That gap is exactly why context match was invented.
The 100% Match Problem Nobody Talks About
Here’s a simple example that shows why a 100% match isn’t always enough.
Say the word “Run” appears in your document. Your TM has seen that word before and saved a translation for it.
But wait. “Run” from a software manual means “execute the program.” “Run” from a fitness app means “go jogging.” These have completely different meanings. The word is identical. The translation should not be.
A 100% match would just grab the saved translation and hand it over. No questions asked. The tool doesn’t care which meaning applies here. It only looked at the word.
A context match catches this. It looks at the sentence before “Run.” If the previous sentence says “To start the application,” the context is clear. That’s software language. The tool matches both the segment and its surroundings before committing to the saved translation.
That small extra step prevents a lot of embarrassing mistakes.
What a Context Match Actually Checks
A context match must pass three tests — all three, not just one.
First: The segment itself must be character-for-character identical to what’s stored in the TM. No exceptions. One different comma and it drops to 100% or below.
Second: The document structure must match. Is this sentence sitting inside a heading? A table cell? A bulleted list? If the structure is different from last time, it’s not a context match.
Third: The segment right before the current one must also be identical to what was stored with it in the TM. This is the real secret sauce. This is the check that catches the “Run” problem.
All three must pass. If any one of them fails, the score drops. It might still be a 100% match, or a high fuzzy match — but it won’t be a context match anymore.
101% vs 102%: What’s the Difference?
People sometimes see both of these scores and wonder why there are two.
The answer comes from how many of the surrounding segments matched.
A 101% match means the segment itself is identical, plus either the segment before OR the segment after also matches. One neighbour confirmed.
A 102% match means the segment is identical, plus the segment before AND the segment after both matches. Both neighbours confirmed.
A 102% is even stronger confidence than a 101%. Both are called context matches, but the 102% is slightly more airtight.
Different CAT tools handle this in slightly different ways. Some tools only use 101% and call anything higher a “perfect match.” But the idea behind all of them is the same — the more context that matches, the more trustworthy the saved translation is.
What Is an ICE Match?
If you use memoQ, you might never see the phrase “context match” at all. Instead, you’ll see an ICE match.
ICE stands for In-Context Exact match. It is the exact same concept with a different name. The tool finds a segment that is textually perfect AND sits in the same context as before.
Trados Studio calls it a context match. memoQ calls it an ICE match. Phrase (formerly Memsource) and Smartcat have their own labels too.
Same idea, different brand name. Avoid becoming confused by the jargon.
Why Context Match Is Such a Big Deal for Translators
Here’s the part that actually matters in your daily work.
When a segment comes up as a context match, you can almost always trust it. The translation is correct, and it fits. You don’t need to second-guess it. You don’t need to spend twenty minutes verifying it.
In practice, project managers often lock context match segments before translation even begins. This means translators never even open those segments. They’re automatically pre-filled and confirmed.
For a large document — say a 50,000-word software manual that gets updated every few months — this is a game changer. Huge chunks of unchanged text just pass straight through. The translator only touches the new stuff.
Jobs with strong context match rates can run 30–50% faster than those without. That’s not a small time saving. That’s the difference between a one-week turnaround and a two-day turnaround.
What It Means for Money
Translation is priced per word. Everyone knows that. But the rate you pay isn’t the same for every word.
Context match segments are usually billed at 0–10% of the standard new-word rate. Some clients pay nothing at all for them. The reason is simple — the translator barely touched those segments. They’re essentially free from a labour standpoint.
Compare that to a fuzzy match at, say, 75%. That might be billed at 40–60% of the full rate because the translator had to read it, understand what changed, and edit it carefully.
And a brand-new word — one the TM has never seen — gets billed at 100%.
So for a company that updates the same product manual every three months, a well-maintained TM filled with context matches can slash translation bills significantly over time. The TM grows smarter with every project. Each confirmed segment becomes a future asset.
Where Context Match Really Shines: Real-World Uses
Software Localization: Every few weeks, a software program may change its buttons, menus, and error messages. Most of those strings haven’t changed. Context match lets translators skip everything identical and only work on the new text.
Legal Documents: Contracts and legal clauses get updated often but rarely change entirely. Context matching ensures the unchanged clauses stay precisely translated — not slightly rephrased by accident.
Medical Manuals: In medical translation, consistency isn’t just nice — it’s required by law in many countries. A device that said “press the button firmly” last year should say exactly the same thing this year. Context match enforces that.
E-commerce Product Descriptions: Big online stores with thousands of products often repeat the same phrases — “Free returns within 30 days,” for example. Context match reuses those translations across millions of listings instantly.
The Tools That Use It
Almost every major CAT tool supports context matching. Here’s a quick look at how the big ones handle it.
Trados Studio — The oldest and most widely used. Context matching is on by default. You can see match details in the Translation Results window. Perfect Match (which compares against full bilingual files) is a separate feature.
memoQ — Uses the ICE match label. You can configure it to lock ICE segments automatically during pre-translation. It also supports 102% double-context matches.
Phrase (Memsource) — Shows context matches in the analysis report. Uses metadata called “context keys” to store segment positioning.
Smartcat — Stores not just the source and target text but also the previous and following segments alongside every entry. When both neighbours match, you get 102%. When only one does, it’s 101%.
When Context Match Isn’t Enough — The Human Factor
Here’s something the software won’t tell you.
Even a 101% match can be wrong sometimes. Not because the tool made a mistake — but because human decisions changed after the original translation was saved.
What if the original translator made a small error? That typo or awkward phrasing is now sitting in your TM with full context match status. The tool will trust it completely. The tool has no idea it’s wrong.
What if your brand updated its terminology last year? The old word might still live in the TM. The tool finds a beautiful context match — but with outdated language.
This is why experienced project managers never disable human review entirely, even for context match segments. They might do a lighter review — a quick scan rather than a full proofread — but they don’t skip it completely.
The software is a brilliant assistant. It is not a replacement for professional judgment.
How to Get More Context Matches on Your Projects
Getting a high rate of context matches isn’t luck. It’s planning.
Use consistent document templates. When the structure of your files stays the same, the CAT tool can compare new versions to old ones properly. Change the template and you lose structural context — which drops context matches to plain 100% matches.
Keep your TM clean. Remove old, incorrect, or outdated translations regularly. A messy TM gives you confident wrong answers.
Standardize terminology. Use a termbase (glossary) so the same word is always translated the same way across projects. Inconsistent terminology breaks context over time.
Avoid unnecessary rewrites. If a paragraph doesn’t need to change between document versions, don’t change it. Every edit resets the segment and costs you a potential context match in the future.
Plan at the project level. Project managers who lock context match segments before work begins can reduce translator workload by 20–35% on large content-heavy jobs.
The Future: AI and Smarter Context Matching
The current system is rule-based. The sentence must be character-for-character identical. The surrounding segment must be exactly the same. It’s strict, and that strictness is part of what makes it reliable.
But researchers are already working on something more flexible — semantic context matching. Instead of checking whether the previous sentence is word-for-word identical, future tools might check whether it means the same thing, even if the wording changed slightly.
That would let a CAT tool recognise, for example, that “Press the green button” and “Click the green button” are functionally the same context for the sentence that follows. Today, those would count as different contexts. Tomorrow, they might count as one.
This shift is still in the research stage. For now, the 101% match you see in Trados or the ICE match in memoQ is still a rule-based, exact-context check. But the direction is clear.
The tools will get smarter. The translators who build rich, well-maintained translation memories today will benefit the most when that next generation of matching arrives.
Final Words
Context match sounds like a technical term. And it is. But the idea behind it is beautifully simple.
Words only make sense in context. “Run” means something different in a gym app than in a server manual. “Free” means something different in a pricing table than in a welcome message. Language is always surrounded by other languages — and that surrounding language shapes meaning.
Context match is the translation industry’s way of respecting that truth. It doesn’t just ask “have we seen this sentence before?” It asks “have we seen this sentence, in this exact situation, with these exact neighbours?”
That’s a smarter question. And smarter questions lead to better translations.
If you work with translation regularly — whether you’re a freelance translator, a localization manager, or a company that ships documents in multiple languages — understanding context match puts you ahead. You’ll know why that 101% number isn’t a typo. You’ll know why your TM is worth protecting and growing. And you’ll know when to trust the machine — and when to still read carefully with your own eyes.
FAQs
1. What is a context match in simple words?
A context match means the software found your sentence in its database, and everything around that sentence also matches exactly. Words, position, and neighbours — all three are confirmed.
2. Why does a context match score 101% instead of 100%?
The extra 1% represents the confirmed context — the surrounding sentences. It’s not a mistake. It’s the tool telling you: “This isn’t just a word match. This is a position match too.”
3. Is a context match the same as an ICE match?
Yes. ICE stands for In-Context Exact match. It’s the label memoQ uses for what Trados calls a context match. Both mean the same thing.
4. What is the difference between 101% and 102%?
A 101% match means either the segment before OR after matched. A 102% means both the before AND after matched. A 102% gives slightly more confidence.
5. Can I trust a context match 100% of the time?
Almost always, yes. But not blindly. If the original translation in your TM had an error, the context match will repeat that error confidently. Always do at least a light human review.
6. What happens if I change my document template?
Changing the structure can break context matches. The tool uses document structure as part of its check. A new template means new structural context — so old 101% matches may drop to 100%.
7. Do I still pay for context match segments?
Most clients pay 0–10% of the standard word rate for context match segments. Some pay nothing at all. It depends on your agreement with your translation provider.
8. What is a fuzzy match compared to a context match?
A fuzzy match means the text is similar but not identical — maybe one word changed. Context match means everything is identical including the context. They have very different confidence levels.
9. How do project managers use context match segments?
Most PMs lock them during pre-translation. That means the translator never even opens those segments — they’re automatically filled and confirmed before work begins.
10. Does every CAT tool support context matching?
Most major tools do — Trados, memoQ, Phrase, Smartcat all support it. But the terminology and configuration options vary. Always check that context storage is enabled in your TM settings.
11. Can context matching work with machine translation?
Not yet in the traditional sense. Current context matching is rule-based and exact. AI-driven semantic matching is being researched but isn’t commercially standard yet.
12. How can I increase my context match rate on future projects?
Use consistent templates, keep your TM clean and up to date, standardize terminology with a glossary, and avoid unnecessary changes between document versions. These habits compound over time.
Keep creating, innovating, and inspiring with Content Ideators every day.
