20 March 2026
Why translators sometimes start by deleting the pre-filled target
Yes, it’s true. Sometimes, when translators open up a file in their CAT tool and find it populated with pre-translation, they take one look, then select the target column and press – DELETE.
More and more often, the files clients want us to work on come already pre-translated. They have been pre-populated by AI- or MT generated translation and with legacy matches from their TM. Instructions call for post-editing or lite post-editing. Some of the time, this introduces an Undo action into our workflow:
We delete the entire target column and then start with a clean slate.
But what’s the point, you may ask. It seems totally counterintuitive and wasteful. Why re-do everything when clients are actually doing part of the work?
Why bin it when you could post-edit?
Because sometimes the fastest and safest way to good-quality text is not to edit what already exists, but to step outside it entirely, and start with a tabula rasa.
The tell-tale signs are: pre-translation was done without attaching a glossary of the key terms; non-translatables such as product or company names were mistakenly translated; there is a mix-up between formal and informal forms of address, etc.
Machine translation and TM-based pre-translation are powerful tools. Used well, they save time and ensure consistency. And recycling surely is best practice – not just for clothes and materials, but also for content.
Yet, looked at it from a practitioner’s angle, it does introduce unwanted effects.
Among them is the fact that once a target text exists and is proposed by the CAT tool, human translators, reviewers or post-editors naturally tend to work within that frame. Yes, they change the odd word here and there, add a bit of pizzazz, or a few commas or n-dashes, choose a more idiomatic adjective. But, all in all, they stay close to what’s already there.
Experience proves that copywriters and translators are less inclined to demolish a structure that already stands, even if it was built for a different context, a different brand voice, a different audience, or a different moment in time. It’s comforting to re-use building blocks or to get a 100% match and just press “Confirm”.
The result is often acceptable and faithful, but slightly stiff, or lacking cohesion. It doesn’t quite flow, or worse, it does not encourage the reader to continue reading and engage.
Language evolves over time, and marketing copy is not written in the same way it was 20 years ago. The same is true even of operator instructions. These days, they tend to use a less stiff and formal language, simpler syntax, shorter sentences, and fewer words. The interaction with consumers has become less formal, more empathic, more personalised. Users expect not only new content, but some entertainment value – and that includes stylistic variation. In many cultures these days, they also expect gender-neutral, inclusive language.
In many languages, it involves the change from formal to informal tone (“vous” – “tu”, “Sie” -> “du”). Or a new way of asking people to get in touch to find out more (cf. “Wenn Sie Fragen haben oder weitere detaillierte Auskünfte wünschen, stehen wir Ihnen jederzeit gerne zur Verfügung“ à „Kontaktieren Sie uns“; „Sehr geehrte Damen und Herren“ à „Guten Tag“, or even „Hallo“).
We see legacy phrasing resurfacing where it no longer fits. In user guides, in white papers, in marketing copy. English syntax bleeds into the target language. Messaging starts to feel dated and predictable, even when the product is new. Over time, these details accumulate.
Under these circumstances, deleting the target column is sometimes a natural impulse for stopping that drift.
When we remove pre-existing target text, we create space for a different workflow.
Instead of tweaking and repairing segments, we look at the text as a whole. Instead of asking “how can this be improved?”, we ask “how do I write/translate this for this audience, right now?”
Modern AI systems are good at generating fluent, coherent drafts at the text level. Sometimes uncannily so. By now, everyone is aware that they are fallible and cannot be trusted 100% (just like humans). But with proper guidance and care, they can establish a clear structure, use the proper register, and create flowing, readable text (or translation) without being anchored to outdated phrasing, sentence-by-sentence alignment, or reliance on existing material.
The writer/translator then steps in, not as a repair technician or a cleaner, but as an editor, strategist, and final decision-maker. Tone is adjusted. Terminology is safeguarded. Risk is evaluated beforehand. Cultural nuance is applied. Superfluous repetitions do not occur. A nice little extra phrase may be added, for better understanding and readability.
Paradoxically, this often results in less human effort, not more. The time is spent where it matters, on the creative side, rather than on endless micro-corrections.
Of course, we do not delete every time. For highly regulated content, stable technical documentation, or environments where strict traceability and consistency are required, working within CAT tools and reusing existing segments remains essential. There is no denying that.
So, deleting the target column is not a rejection of CAT tools. It’s a conscious choice, applied selectively, when the content calls for freshness, adaptability, or a stronger narrative voice. Or, of course, when the source and/or the target is of poor quality.
The real question isn’t whether machine translation, raw AI, or CAT tools are “good” or “bad”.
It’s whether the workflow serves the purpose of the content and scenario.
Sometimes reuse is exactly what’s needed. And sometimes, the most professional choice is to acknowledge that the legacy material no longer provides the best blueprint for what is needed now.
Deleting the target column isn’t an act of rebellion. It’s an editorial decision.
About the author
Isabelle Weiss, founder of Alpha CRC, has been a leading voice in the translation industry for almost forty years. Weiss has been a consistent presence in the company since our beginnings in 1987, and continues to work directly with translations for clients.
