Should You Base Translation Quality on Accuracy or Preference?

Should we completely remove preference from translation quaity?


Beauty is in the eye of the beholder. I've always had a hard time grasping the wording of the idiom, but never the message. Beauty is subjective, and we all determine our own definition of beauty. And in effect, beauty is synonymous with quality.

The quality of our favorite books or films, the quality of the news we read, or even the quality of the digital content we consume daily. Quality is ultimately determined by the individual viewer, which begs the question, how do we create quality content, and how do we determine when we've managed to hit that mark?

This is a question localization professionals wrangle with every day: "how do we determine a high-quality translation?"

We've talked a bit before about the idea of quality, how we define the concept, and whether or not it even matters. While there's a lot to unpack there, I think Kate it the nail on the head with an interesting starting point for the conversation:

When determining the quality of translation, what matters more: individual preference or complete and total accuracy?

Understand the intent of your content

Quality is subjective, we know that. Every person sees the world differently, influenced by different experiences, locations, and so much more. Laura said it beautifully, "we all have an individual and unique perspective that shapes how we see, understand, and absorb content."

But, we can of course work to define quality create some guidelines. During Episode 3 of Translation isn't blind, Kate explained that one simple way to determine how you want to define quality should start at the original inception of the content.

Think about "how much time and attention did you put into your source content? What did you think about when you were putting this together? What were the resources used, what was the budget?" These questions help you started around thinking about the end goal of your content.

Ultimately you want to come back to the question: what was your content meant to do?

Which is a really important question to ask around quality specifically? Was your content meant to supplement users and guide them through your app? Or is it critical healthcare information that needs to be accurate?

How can we define quality?

That's how we can start thinking about defining quality, preference, or accuracy?

For example, if your content was created for legal purposes, then accuracy is paramount. You can't have any missing punctuation, typos, or grammatical errors.


Ensuring users can understand the intended message and can do what they set out to (can they track their order or navigate your website?)

Meanwhile, if your content was meant to direct users through your website, then a missing comma might not matter so much; instead, you want to ensure that the user can do what they set out to. This might come down to the preference of individual project managers or translators, or reviewers and editors.


Could be opinion based on voice or tone, word choice, or even preference around grammar (think: oxford comma)

And that might make you stop and think at first. But look at it this way, maybe you're translating a shipment email to your customers if a small typo or missing punctuation, can they still track their orders?

Is preference the right metric for quality?

We're of the position that teams should remove opinion from their translation review at all costs. There needs to be a fine balance between individual preference and constant revisions.

At the end of the day, language is subjective. I wrote before that twenty different people can read the same line, and pull twenty or more different meanings and interpretations. Therefore the goal of translation review isn't to input an individual interpretation into the review, but rather objectively evaluate the quality of the translation.

Which then neatly introduces us to another question, how do we quantify quality to better determine the right path forward?

How do we quantify quality?

I've talked about this a bit before, around how we can determine the value of our content. And a lot of what was said applies here as well.

Quality requirements can also come down to the overall value of your content, again looking back at the end-goal. If it’s a major marketing or sales campaign, then it might drive quite a bit of value for the business. But if it’s a quick patch-note email that only a few select users will read, the value isn't nearly as high.

That's important to consider when asking yourself what matters more, accuracy, or preference. When it comes down to it, accuracy might require a manual internal review step, adding to the time and cost of your project and lowering the value of your ROI. But if the goal is to move product fast, or spread awareness, it makes sense to focus on the right messaging over perfect detail.

Think about what exactly your content is set out to do, and how it will deliver value back to your brand:

  • Are you more concerned about increasing website views?
  • Will you be focusing on translating your most important products?
  • Is this a new market in need of translation, or supplement content?
  • Can you leverage internal resources like Translation Memory to aid translators?
  • Can this be translated and keep its original meaning?

This is where we'll turn to data. For example, we can analyze website traffic to find your most popular products or pages, and dive into GDN requests to understand where your viewers and users are coming from.

Accuracy vs Preference: Which is right?

It's hard to know you're looking at the same thing and someone can see something different, and that's a big challenge in translation.

But we can establish a solid middle ground by defining quality, and determining the value of our content.

The end goal is to find that balance between accurate, correct, and error-free content, while also delivering back that value that was initially invested, and accomplishing the original intent for every project.