Field Notes

AI and Localization in Progress. Things are changing fast for people in the localization world. This podcast from features short 15-minute conversations with industry thought leaders to keep you up to date on the latest innovations, experiments, and challenges.

All Episodes

Field Notes

Are Fuzzy Matches Dead?

May 14, 2026 • Stephanie • Episode 2

0:00 | 12:21

Fuzzy matches have been treated like a law of physics in localization: higher similarity means lower effort, so the discount grid must be fair. But when we look closely, that assumption starts to wobble. Steph sits down with Erik Vogt to ask the blunt question many teams are now debating out loud: are fuzzy matches dead, or are we just finally admitting we never proved they saved time the way we claimed?

We unpack how machine translation and modern AI review workflows challenge the entire “segment similarity equals effort” model. Erik argues that even 100% matches can demand real validation when context shifts, and that linguists increasingly need tools that surface accuracy risk rather than a fuzzy percentage. We talk about quality estimation (QE), MTQE, and LQA signals, and why QA is evolving from basic checks into accuracy-focused guidance that helps humans get to the right answer faster.

Then we go a step further into where LangOps may be headed: object-based translation. Instead of translating line by line, AI can rewrite an entire asset as a cohesive object and tune tone, reading level, and intent, which raises big questions about repetitions, translation memory, and word-count pricing. We close by reframing the center of localization as human-in-the-loop validation and authenticity, not score-driven leverage.

Subscribe for more conversations on AI in localization, translation memory strategy, MT quality, and the future of language operations, and if this sparks a strong opinion, share the episode and leave a review.

Are Fuzzy Matches Dead?

Stephanie Harris-Yee 0:06

Hi, I'm here with Erik Vogt, and we are going to be diving into the question that seems to be popping up more and more recently, and that is, are fuzzy matches

How Fuzzy Matches Shaped Pricing

Stephanie Harris-Yee 0:16

dead? So this is of course talking about the concept of when you're doing a translation and you have a translation memory, the system will remember that you have translated something before. And if a group of words is similar in new the new content, but not exact, the system will flag that as a fuzzy match, which are supposedly much quicker, easier for a translator to validate, and thus should come with a discount. Now, Erik, why could this whole concept, this whole system, be going away?

Erik Vogt 0:47

Steph, that's it's really interesting that it even exists in the first place, honestly, because I don't think it's ever been actually validated that fuzzy matches are by definition easier to work on than not. One can even make the argument that 100% matches aren't necessarily easier to validate than not. Just because they are match something that exists doesn't necessarily mean that they've that what you're leveraging it from is appropriate for what you're leveraging to. But in the industry, we've set up this arc, this framework, this this heurist for time savings, a discount savings based on similarness. And I think that AI, MT and AI both have challenged that. So I think in even when MT was first coming out, there was a question as is MT better or is the fuzzy match better when you're comparing apples to apples, which is going to get to the closer outcome? The fuzzy match is already, you already see that it's not as it's not perfect. You have to

MT And AI Start Replacing Fuzzies

Erik Vogt 1:47

work on it. MT, you may not have to work on it. It may be more accurate than the fuzzy match. So then you think maybe it depends on how big the gap is. So we talk about everything below 75% match, for example. Let's do MT on that. Or maybe it's everything 85%. Will you MT getting better and better? Then we would presume that the MT would chew up more and more of this fuzzy match categories to the point where there are no longer any question about it that the MT is going to be better than fuzzy matches in all cases. And now this is MT that we're talking about. Now we're moving into an AI review paradigm, like the one we have with Mosaic. We're not the only ones, but we have different capabilities here that AI is already taking the fuzzy matches and adapting them to what it thinks the correct answer should be. And it's the it's interestingly that the UI then, the translator task now is not to work at a different rate depending on whether it's a low or high or medium fuzzy, which is remember that's the implicit assumption, right? That it's more it's easier for a translator to fix something that has a high fuzzy score than it is for a low fuzzy score, which again is an for the most part an untested hypothesis, but in general we accept as kind of a standard in our industry as a heuristic. Now the reviewers are specializing more in getting the final content correct, and they don't see any difference between what the fuzzy match originally was. In fact, you might be looking at totally different signals, such as an LQA score or a QE score or other sort of tools that you're using to flag things that might be wrong for different reasons. The QA tools now are shifting from being things like spell checkers to being things like accuracy checkers, because that is now more important. The most important thing is that you is the humans contributing an awareness of what the right answer is. So again, we're thinking about this task here where a linguist is trying to like have easy access to all the tools that they can get to to come up with a correct answer, but its fuzzy score may not be that useful of a tool anymore. And I think arguably, like I said, there's folks who have been saying MT already may have been there before AI came along. But now with AI doing this final step here, maybe this isn't really about fuzzy. Maybe fuzzy matches are just an extra layer of complexity that we're adding to our quoting that isn't really quantifiable in terms of the value that it's creating for the translators. And let's be honest, like we want the

Measuring Effort And Protecting Accuracy

Erik Vogt 4:35

translator to be successful. Like we want to have them be effective at what they're doing. Another, just as a minor segue on this, which is a related point, there the a lot of the technology providers in the industry make technology to try to facilitate the work of the translator, and on the flip side, to drive a hypothetical cost reduction, right? So you can create this technology and you say it's gonna save 20%, and then you put it into the wild and it may or may not deliver 20%. You can pay 20% less or get paid 20% less, but it doesn't necessarily mean that the work has reduced by 20%, mainly because we don't really check effort, particularly in our industry. We check outcome. So I think that as we switch to a more and Lilt kind of tackled this, and I think RWS also had this adapted MT model that was designed around this principle that you are focused on a KPI, which is words per hour. Like you're how fast are you working? And you provide that feedback to the translator and they can see that how they're doing, and you're basically in scenting speed. We know that speed is inversely correlated with accuracy. So these are countervailing kind of trends. But suffice to say, ultimately that really matters is time. And what really matters to the translator is helping them to get things right. So I'm not sure anymore that fuzzy matches are doing that much to help translators do their job, when especially when you have so many other powerful tools out there like machine translation, which by the way, more often than not, now is enhanced with AI technology on the back end. So MT is getting better all the time. And we have extra AI quality layers that are being added on top of it that are now enhancing even more. And of course, we've talked about LQA and MTQE and other technologies that are designed to analyze input and establish whether or not it's so how many different of these tools do we need? I would say maybe fuzzy matches really are on their way out, and we don't really need these anymore. And we should mainly, and this is something I believe in my heart, is that we should keep reminding ourselves that at the center of this whole story of localization is the enhancement that the human is providing, the validation and the authenticity of their knowledge and their validating that this is true, whatever this segment is true and correct for this output. So, anyway, these are just a bunch of thoughts on this. Lastly, I'll just close by saying one other

Object Based Translation Changes Everything

Erik Vogt 7:18

thing. There's there's an argument that could be made that 100% matches and repetitions also should be threatened in this new world as a model, largely because if you think about a segment-based model that is breaking down a project into words that presume that this source equals this target, you at best with exact matches or looking above like above and below the line to see is are is what are its neighbors also the same, then I'm going to be more confident that this is exactly the right because I I believe that what we did in the past is exactly right. But we also have this capability to do a summary of an entire object and to have AI retranslate that entire object as an entirely new object. It's not, it's no longer a word-based model, it's an object-based model. And now you can tune up or down that output based on totally different parameters. So you could take the English article that is about 10 ways you're screwing up your coffee and turn it into a Chinese article that's 10 ways to delight your family with excellent coffee or tea, right? You can change those things tone-wise, you can change it in terms of vocabulary, you could tune it down to a fifth grader level or up to a PhD level. There's so many things you can do now at a holistic level. And I'll do a call out to those in the Lang Ops space who are, I think, very interested in exploring this object-based model, getting away from word count model. In any case, my hypothesis is that this time it might be for real. Maybe fuzzy matches really are on their way out, and we're going to see a new uh evolution towards a model that it completely excludes them and focuses on the tools and controls that we have that are deeply connected to making the human in a loop the most valuable as we possibly can make them.

Stephanie Harris-Yee 9:14

Okay.

Translation Memory As Training Data

Stephanie Harris-Yee 9:15

Here's maybe a tangential question. Do you see this then affecting the value of translation memory in general? Or is that still going to be super important for folks to make sure that it's clean, it's ready to go, just for like custom training the MT side of things?

Erik Vogt 9:31

Okay. There, there's two different use cases for TMs there. And one is as a leveraging tool, and one of them is a training tool. And I think you're right, there are already a significant number of deployments in which MT TMs have only been built for the purposes of training an MT. So, and this is particularly in kind of high-volume e-commerce contexts where you're just trying to get to good enough to make a buying decision type of level look of quality. But I think that the the this does illustrate an excellent point that what you're what you need to train your AI, if you think about completely unpacking the system and starting over again, might be totally different than what we're used to. So we presume that language layer is really just source and target segments, but maybe there's things like knowledge graphs and product tables that define the characteristics of the objects that you're trying to describe. And then, of course, there's the glossaries, which are mission critical. There's things like style guides, which are mission critical, but the TM itself is a byproduct of the application of glossary and uh and style guides. And if you can properly train MT or an AI with the appropriate guidance to replicate that, then yeah, a TM may not be something you need to use day to day as a leveraging tool. I suspect it will still be useful as a training tool. So if you've bothered to pay a human to review and you still care about that segment-to-segment parity, then that can still be a training asset for an MT or AI, which to be honest, they're not 100% different concepts. Like neural MT and AI are basically very similar core technologies, but there's very different applications of these things. And as we attack our localization process of the future, I 100% think that we need to question the value of an MT as well, especially old big ones that require a lot of cleanup. We do provide that service that we it is possible, but we see a lot of times the reason for that cleanup is to train an MT, not to keep it in a leveraging function forever.