Localisation 10 August 2026
Fit-for-Purpose Quality: What Does Good Enough Mean in Localisation?
Not every piece of content requires the highest quality. The fit-for-purpose quality approach aims for the best result within the constraints that exist. We look at thresholds and quality tiers.
There is a sentence you often hear when a localisation budget is being discussed: “We want the best.” The intention is understandable, but on its own the sentence does not produce a criterion. An organisation that invests at the same level in all its content usually spends on ordinary content the resource it should have kept for critical content.
Contents
The literature has a name for this problem: fit-for-purpose quality. Miguel A. Jiménez-Crespo notes that in this paradigm the aim of quality evaluation is to reach the best possible degree of fitness for purpose within the constraints that exist (2024, pp. 62-63).
This article deals with the concept of fit-for-purpose quality, the necessary and sufficient rule, quality thresholds, the quality tiers that vary with the type of content, and how the user perceives quality.
What is fit-for-purpose quality?
Rather than aiming at absolute quality in every case, the fitness-for-purpose paradigm proposes looking for the best degree of fitness for the purposes that have been set. In doing so it also takes account of the constraints that exist: financial resources, time, human resources and procedural limits (Bowker 2019, cited in Jiménez-Crespo 2024, pp. 62-63).
The critical distinction in this approach is that purpose and resource are considered together. A model that seeks a perfect result with an unlimited budget does not work in real projects. Any definition of quality that leaves constraints outside the equation stays on paper in practice and is of no use at all at the moment of decision.
It is also a concept open to misunderstanding. Fitness for purpose is not a polite name for low quality. For critical content, the fit-for-purpose level is the highest level. The only thing the approach says is that not every piece of content requires the same level.
The approach also turns the quality discussion from a negotiation between provider and client into a shared plan. When purpose, constraint and expected level are written down together, both sides know what they are committing to. Most of the disagreements that arise later come from these three pieces of information never having been discussed.
The necessary and sufficient rule
In this paradigm, quality evaluation follows the “necessary and sufficient” rule (Wright 2006, cited in Jiménez-Crespo 2024, p. 63). That is, the question asked is not “could this be better” but “has the level needed to reach the purpose been reached”.
A sufficient degree of fitness can also be established through user-based evaluation methods. The example in the book is very concrete: users can be asked whether they were able to solve the problem that brought them to an FAQ page. If the answer is yes, the content counts as fit for its purpose.
This evaluation can hold even if there are small errors or shortcomings, and even if the content was localised with a machine translation system. In some cases, fitness for purpose means accepting that a small number of errors is unavoidable (Jiménez-Crespo 2024, p. 63).
How are quality thresholds set?
One of the ways the paradigm is applied in the industry is through “quality thresholds” (O’Brien 2012a, cited in Jiménez-Crespo 2024, p. 63). A threshold defines the limit of error weight that must not be exceeded for a piece of work to count as acceptable.
Thresholds can differ even within a single organisation. Thresholds can be defined that vary by product, content type, project and context. Using the same threshold for a company’s contract texts and its product descriptions usually leads to a waste of resources.
For a threshold to work, two conditions must be met: the error classification and the severity scale have to be defined in advance. Without these two, a threshold is no more than an arbitrary line drawn on a scale nobody has agreed on.
The soundest way to set a threshold is to draw on data from past projects. Measurements are taken on the first two or three projects and averaged, and the threshold is then set slightly below that average. Thresholds set with no data at all turn out either too tight to be reachable or too loose to filter anything out.
Quality tiers: matching content type with process type
The book presents a continuum that relates quality levels to content and process types (Jiménez-Crespo 2024, p. 63). On this continuum, high quality is associated with professional human translation and with critical content types. The table below summarises that match in simplified form.
| Expected level | Content type | Typical process |
|---|---|---|
| Very high | Law, medicine, international relations | Professional human translation |
| High | Advertising, games, literature | Professional human translation |
| High | Software, interfaces, business applications | Machine translation with professional post-editing |
| Medium | General content, website copy | Light post-editing by a professional |
| Low | User-generated, short-lived content | Raw machine translation |
The reasoning behind the top level does not appear in the table, but it matters. The book notes that the highest quality is also sought where a person’s life could be put at risk: instructions on how to repair the brakes of a vehicle, say, or how to use a pacemaker after heart surgery (2024, p. 63).
Is a low-quality translation better than no translation at all?
This is the most uncomfortable question in the debate. The book reports that in certain situations the industry accepts the following view: offering some translations with the resources available is better than offering none, even if the quality is lower (Drugan 2013, cited in Jiménez-Crespo 2024, p. 63).
The areas where this view holds are limited. Short-lived content, user-generated texts and content read only to get a general idea fall into this group. The same reasoning cannot be used for critical content.
The question to ask when making the decision is this: what happens if this content is misunderstood? If the answer is “the user will struggle a little”, flexibility is possible. If the answer is “the user could be harmed” or “the organisation could face legal liability”, there is no flexibility.
This decision also has a communication side. Marking content that has been localised to a low level as such aligns expectations from the outset. A help article stated to have been translated automatically draws far less negative reaction than the same text published silently.
How do users perceive quality?
The user’s perception works differently from the expert’s evaluation. Users do not count errors; they look at whether the content did the job for them. This is why the fitness-for-purpose approach accepts asking users directly as a legitimate method of measurement.
Users do, however, notice some errors very clearly. The book reminds us that users and fans frequently comment on errors in games or on websites (Jiménez-Crespo 2024, p. 63). In other words, flexibility quickly reaches its limit in highly visible texts.
The practical consequence is that content has to be separated according to its visibility. The home page headline, the payment screen and error messages are highly visible; archive pages and rarely read help articles are of low visibility. Distributing the quality investment according to this map gives a better result on the same budget.
How is fitness for purpose measured?
Measurement calls for indicators that are independent of error counts. The list below shows a simple measurement set that most organisations could build from the data they already hold.
- Task completion rate: can the user finish what they came to do?
- Resolution rate in help content: did the page solve the problem?
- Number of support requests: has it risen or fallen since localisation?
- Cart abandonment or form drop-off rate.
- How often language-related complaints appear in user reviews.
None of these indicators is sufficient on its own, but read together they give a more reliable picture of fitness for purpose than error lists do. What is more, most organisations already collect this data and never use it in localisation decisions. Comparing the values before and after localisation often does not even require setting up an additional measurement.
The Türkiye context
In Türkiye, translation purchasing decisions are mostly made on a single unit price and all content is gathered under one heading. Yet the contract annexes and the campaign announcements of the same organisation carry different risks. Separating content into tiers can raise the quality of critical texts without increasing the total budget.
The fact that Turkish is a relatively low-resource language also has to be taken into account. While machine translation output gives acceptable results in some fields, it can weaken markedly in technical fields where the terminology has not settled. For that reason it is safer to take the tier decision field by field.
On the regulatory side there is little room for flexibility. Consumer information, instructions for use of health products and financial contract texts are among the content types where the margin of error has to be kept to a minimum. This article is for general information only and is not legal advice.
What does this mean for you?
The fit-for-purpose quality approach is a decision tool that lets you distribute your budget more intelligently. Applying it does not require you to build expensive infrastructure; the only thing you have to do is stop seeing your content as a single mass and assess the risk each part carries separately.
Separate your content into three along the axes of risk and visibility: critical, important and ordinary. Define a different process and a different threshold for each tier. Use a professional process and strict checking in the critical tier and a lighter flow in the ordinary tier. Making this distinction once means it can be reused in every project that follows.
Once you have made that distinction, you can compare which service level suits which tier on our translation services page, and use our quotation form for a tier-based budget. In most organisations a tiered approach raises the quality of critical content without increasing the total cost.
Frequently asked questions
Is fit-for-purpose quality a polite name for low quality?
No. Fitness for purpose targets the level needed to reach the purpose within the constraints that exist. For critical content, that level is the highest one. The only thing the approach says is that not every piece of content requires the same level.
What is a quality threshold?
It is the limit of error weight that must not be exceeded for a piece of work to count as acceptable. Thresholds can differ by product, content type, project and context. Different thresholds can be defined within the same organisation for marketing copy and for help content.
Which type of content requires the highest quality?
Critical content such as law, medicine and international relations, along with creative texts such as advertising, games and literature, require the top level. Content that could put a person’s life at risk or damage an organisation’s reputation also falls into this group.
Is a low-quality translation better than no translation at all?
The literature reports that under certain conditions this view is accepted: offering some translation with the resources available may be better than offering none, even if the quality is lower. This assessment does not hold for critical content.
Sources
- Jiménez-Crespo, M. A. (2024). Localization in Translation. Routledge, pp. 62-63.
- Bowker, L. (2019). The fitness-for-purpose paradigm; cited in Jiménez-Crespo 2024, p. 62.
- Drugan, J. (2013). Quality in Professional Translation. Bloomsbury, p. 180.
- O’Brien, S. (2012). The concept of quality thresholds; cited in Jiménez-Crespo 2024, p. 63.