Skip to main content
Upcoming discountNext discount: 10% off — today 19:00.starts in--:--:--Get a quote →
Client Login
Home News Translation Outsourcing in the EU Institutions: Why Is Work Placed Outside?

Industry and Profession 5 August 2026

Translation Outsourcing in the EU Institutions: Why Is Work Placed Outside?

The number of in-house translators at the European Commission has fallen by roughly half in twenty years. We look at the reasons behind the shift, the structure of the tenders and the effects on quality.

For many years the institutions of the European Union did their translation with their own staff. Over the past twenty years that picture has changed markedly. Translation outsourcing in the EU institutions has become a permanent method that now covers a significant part of capacity.

Contents

The scale of the change is clear: over a period of roughly twenty years, the number of in-house translators at the European Commission fell by around 50% (Strandvik 2017). This does not mean that the work has shrunk; it means that the work has moved.

This article looks at the reasons for the shift, at how the tender mechanism is built, at the effects on quality, and at the lessons a private-sector buyer can draw from this experience.

How far has translation outsourcing in the EU institutions grown?

Cutting in-house capacity roughly in half is not an easy decision to take inside an institution. A change of that kind usually rests not on a single cause but on several pressures pushing in the same direction.

The first pressure is enlargement. Every new member joining the Union brought a new official language and therefore new language pairs. Building an in-house team for each language ran into limits, both of budget and of the availability of qualified candidates.

The second pressure is budgetary discipline. Limits on headcount in public institutions prevented a growing workload from being met by adding posts. The third pressure is technology: as memory and terminology infrastructure matured, it became easier for translation produced outside to meet the institution’s standard.

Four reasons for placing work outside

Public or private, every organisation that turns to outsourcing shows the same four reasons. The first is that demand fluctuates. When the legislative calendar gets busy, volume rises sharply; a fixed establishment cannot meet those peaks, and if it is enlarged to meet them it sits idle in the quiet periods.

The second is the range of languages. There is a constant flow of work in the main language pairs, while in rare pairs a handful of jobs may come up in a year. Keeping a salaried translator for those pairs cannot be justified economically.

The third is specialisation. Agricultural legislation, competition law and chemicals regulation call for very different bodies of knowledge. The fourth is cost; what is striking, though, is that cost usually comes last on the list. Institutions turn to outsourcing mostly for flexibility rather than for cheapness.

How is the tender mechanism built?

Public procurement of translation usually runs through the framework agreement method. A qualification stage comes first: providers are assessed on their financial standing, their capacity, the qualifications of their translators and their processes. Those that pass this stage are placed on a list.

In the second stage the work is distributed within that list. Distribution is sometimes by ranking and sometimes by a mini-competition held for each job. A provider’s position in the ranking may be updated from period to period according to the assessment scores of the work delivered.

The most important feature of this structure is that price is not the only criterion. Quality score, delivery performance and process compliance all enter the assessment. In a model where the lowest price automatically wins, quality is pulled down quickly; this is one of the well-known risks of public procurement.

Another function of the framework agreement is to provide continuity. When the same provider team works for the same institution over a long period, a body of terminology and style builds up. A structure that puts every job out to tender again may look cheap, but it resets that accumulated knowledge every time.

ElementIn-house capacityOutsourcing
Cost structureFixedVariable
Meeting demand peaksLimitedHigh
Accumulated house languageHighAchieved through process
Confidentiality controlDirectContract-dependent
Quality monitoringContinuousBy sampling

The quality cost of outsourcing, and how it is monitored

Outsourcing does not lower quality by itself; unmonitored outsourcing does. Corporate buyers therefore assess a defined sample of the delivered work at regular intervals. The assessment rests on an error typology: meaning errors, terminology errors, language errors, formatting errors.

Public-sector work on the evaluation of outsourced translation points to two conditions for the system to function (Strandvik 2017). The first is that the criteria are communicated to the provider in advance and in clear terms. The second is that feedback is concrete; instead of saying “quality is inadequate”, the type and the location of the error should be shown.

A third condition is that the outcome carries consequences. If the assessment score does not affect the ranking, the volume of work or the renewal of the contract, the system soon turns into a formality. Measurement changes behaviour only when it leads somewhere.

Assessment also has a hidden cost. Reviewing a sample requires an in-house specialist who knows the language and can set the time aside. Where that resource is not allocated, the monitoring exists only on paper. The most common mistake made by organisations moving to outsourcing is to leave the cost of monitoring out of the savings calculation entirely.

Mandated tools and the effect on the supply chain

Large buyers do not merely place work; they also impose a way of working. The best-known example is the choice of tool: the 1996–97 European Commission tenders were decisive in making Trados the industry standard (Garcia 2015). Once working in a particular file format is made a requirement, the whole chain has to move to that tool.

The cost of that requirement usually stays with the person at the far end of the chain. In one case study, 91% of translators had bought the licence for the tool they used themselves (Pietrzak & Kornacki 2021). A buyer who imposes a tool has indirectly passed this cost on to the translator.

The second effect is compatibility problems. In the same study, 90% of translators reported file compatibility problems, and 70% of those said they ran into them frequently. The more rigid the buyer’s choice of tool, the greater the friction in the chain.

Mature buyers therefore specify a file format rather than a tool brand. Requiring work in standard exchange formats leaves the provider free to choose its own tool while giving the buyer data portability. This is the most practical way out of brand lock-in.

The confidentiality and liability chain

The least discussed dimension of outsourcing is confidentiality. The moment work leaves the institution, the number of people who will see the document increases. If the provider uses a subcontractor of its own, the chain grows by another link.

In corporate practice this risk is managed with three instruments: making the use of subcontractors subject to written approval, extending the confidentiality undertaking across the whole chain, and defining in the contract where the documents will be stored.

The third instrument is becoming steadily more critical, because cloud-based translation platforms blur the physical location of the data. Where an unpublished draft law or an unsigned contract is involved, the country whose servers hold the data is a compliance question in its own right.

A fourth measure is to limit the retention period. A measurable provision such as “all copies are destroyed thirty days after delivery” is far more useful than a phrase like “the utmost care will be taken over confidentiality”. The undertaking that can be audited is the one that can be enforced.

When does outsourcing come back in?

Outsourcing is not a one-way street. Institutions do bring some work back in from time to time. There are three triggers for that decision, and all three are measurable.

The first is that volume becomes steady. When the fluctuation eases and the work becomes predictable, the unit cost of fixed capacity drops below that of external supply. The second is a rise in the confidentiality classification; if certain document types must not leave the building under any circumstances, in-house capacity is unavoidable.

The third is a rising coordination burden. Working with a large number of providers demands serious management effort for terminology approval, feedback and delivery tracking. Once that effort exceeds the cost of a post, outsourcing loses its advantage. For most institutions the right balance is a mixture: a core in-house team, with external capacity for the peaks. You can find a comparison of the models on the translation services page.

The Türkiye context

In Türkiye, public institutions usually procure translation through a service procurement tender. In these purchases, carried out within the framework of the Public Procurement Law, the assessment is in effect made on price whenever the technical specification is not well written. That is the most common problem weakening the quality criterion.

A well-written specification closes much of the gap. Once translator qualifications, a mandatory second reading, delivery of the term list, transfer of the translation memory to the contracting authority and acceptance criteria all appear in the specification, it becomes harder for the lowest price to win on its own.

Where documents contain personal data, a data processor relationship also arises under Personal Data Protection Law no. 6698. There needs to be a written data processing arrangement between the contracting authority and the contractor, the use of subcontractors needs to be made conditional, and it needs to be made clear whether the data will be transferred abroad. This article is for general information and is not legal advice.

What does this mean for you?

What public institutions have learned over twenty years is simple: placing work outside lowers cost, but it also lowers quality unless monitoring is put in place. The same lesson holds for buyers on a far smaller scale.

  1. Write down your quality criteria before you award the work; criteria introduced afterwards only produce arguments.
  2. Assess a regular sample rather than every job; this is both workable and sufficient.
  3. Give feedback with the type and the location of the error, rather than settling for a general comment.
  4. Link the assessment result to how the work is distributed; measurement is only useful when it leads somewhere.
  5. Make the use of subcontractors subject to written approval and extend the confidentiality undertaking across the whole chain.

A buyer who applies these five points can get consistent results without building in-house capacity. If you would like to review your current procurement arrangements, you can share sample documents through our quote page and we can work out the scope and the monitoring plan together.

Frequently asked questions

How much has outsourcing grown in the EU institutions?

Over a period of roughly twenty years, the number of in-house translators at the European Commission fell by around 50% (Strandvik 2017). The difference is largely met by external supply.

Is cost the main reason for outsourcing?

Cost is not the only reason. Fluctuating demand, the sheer number of language pairs and the range of specialist fields are at least as decisive as cost.

How do public translation tenders work?

They usually run through the framework agreement method. Providers are assessed in advance and placed on a list, and the work is distributed within that list by ranking or by scoring.

Does outsourcing lower quality?

Not necessarily. What matters is how the work placed outside is monitored. In systems with regular sample-based assessment and feedback, quality can be maintained.

References

  • Pietrzak, P. & Kornacki, M. (2021). Using CAT Tools in Freelance Translation: Insights from a Case Study. Routledge Focus, pp. 60–92.
  • Strandvik, I. (2017). “Evaluation of Outsourced Translations”. European Commission, DGT.
  • Garcia, I. (2015). “Computer-aided translation”. Routledge Encyclopedia of Translation Technology.
  • Pym, A., Grin, F., Sfreddo, C. & Chan, A. L. J. (2012). The Status of the Translation Profession in the European Union. European Commission, DGT.
  • Personal Data Protection Law no. 6698 (Türkiye).
Share

Document ready? See your price in minutes

Get an Instant Quote
See your price in 2 minutes Get an Instant Quote
Chat on WhatsApp