Digital Products and Accessibility 5 August 2026
What Is Usability? Five Measures of a Localised Product
Usability measures how effectively and efficiently a user reaches their goal. In a localised product, what answers to those measures is the text itself.
An interface text can be flawless in grammatical terms and still fail to do its job. If the user cannot find the button, abandons the form halfway through or has to carry out the same action a second time, there is a quality problem. The field in which that problem is measured is called usability.
Contents
The concept reaches translation from outside; it was developed in design and in human-computer interaction. Even so, it bears directly on localisation, because most of the relationship a user builds with a product runs through words.
This article takes up the definition, the five measures, the notion of error, the mental model the user brings along and the relationship with text quality. The aim is to add a second axis to translation evaluation, alongside linguistic accuracy.
What is usability?
The definition in the international standard runs as follows: the extent to which a system, product or service can be used by specified users to achieve specified goals with effectiveness, efficiency and satisfaction. There is an important qualification at the end of the definition: in a specified context of use.
That qualification tells us there is no absolute measure of usability. The same interface can be efficient for an expert who uses it ten times a day and confusing for someone opening it for the first time. Measurement is always carried out for a particular user and a particular task.
The three key words in the definition are distinct from one another. Effectiveness is whether the goal is reached. Efficiency is how much effort that takes. Satisfaction is whether the user comes away from the process content. A translation can affect all three at once.
In localisation these three measures turn into concrete questions. Did the user reach the page they were after? How many steps did it take? Did they feel out of their depth along the way? These questions are answered only when the product is tried out with real users.
Five measures: the components of usability
Five components are widely accepted in the field of web usability (Nielsen 2012). They are described as quality attributes, and each of them can be measured separately. The table below shows what the components correspond to in translation terms.
| Measure | The question it asks | What it corresponds to in the text |
|---|---|---|
| Learnability | How easy is the basic task on first use? | Labels that are familiar and predictable |
| Efficiency | How fast is it once it has been learned? | Short, unambiguous action names |
| Memorability | Is it still remembered after a period away? | Always using the same word for the same action |
| Satisfaction | Is using it a pleasant experience? | A register whose tone does not blame the user |
| Errors | How many errors are made, and how quickly does the user recover? | The clarity of warning and confirmation text |
These five measures do not always support one another. Abbreviations that are efficient for the expert user can lower learnability for the new one. Which measure comes to the fore has to be decided by looking at the product’s audience.
From a translation point of view, the measure that is easiest to improve is memorability. Even a small glossary that keeps terminology consistent markedly reduces the relearning burden when the user comes back to the product.
Not every measure is tested by the same method. Learnability is measured with new users, memorability with users who have been away for a while. A short questionnaire is enough for satisfaction, whereas efficiency calls for task times to be recorded. When a measurement plan is drawn up, the priority measure has to be chosen at the outset.
Whose error is the error here?
One of the measures is error, and it is often misunderstood. What is meant is not the defects that arise from the development or translation process. It is the wrong actions the user performs while interacting with the product: pressing the wrong button, filling in a form incompletely, going into the wrong menu.
Errors fall into two groups. Minor errors are cases in which the user can easily go back and carry on. Catastrophic errors are those the user cannot recover from at all, or recovers from only with great difficulty. The second group produces direct commercial consequences, such as an abandoned shopping basket.
Text affects both groups. A vague button name increases mistaken clicks; an error message that cannot be understood can make it impossible for the user to get back. Translation is a variable that can change the error rate directly.
This is why the design is stripped back and navigation menus are hidden in critical flows such as purchasing. The aim is to reduce the elements that distract attention and, with them, the likelihood of error. The same plainness should be sought in the text.
The mental model the user brings along
A user opening a new site for the first time does not arrive with an empty mind. They arrive with a stack of expectations accumulated from the thousands of pages they have visited before, and they carry a general model of how home pages ought to work (Nielsen and Tahir 2002).
That model also determines how the user will process the text. Headings such as basket, my account and contact need to be in the expected place and under the expected name. A creative but unfamiliar rendering may look original, yet it slows the user down.
In localisation this calls for a look at the habits of the target market. The literal equivalent of a pattern that is established in the source language may not be established in the target language. The right question is not whether the word is correct, but whether it is the word expected in that market.
The quickest way to answer that question is to look at local products in the same sector. The words competitors use are the de facto state of the model in the user’s mind. A brief survey gives a more reliable starting point than dictionary work.
The same principle holds for software and games. Following platform conventions means preserving the patterns the user already knows. Breaking those patterns deliberately can be defended only where there is a strong reason for it.
Is text quality part of usability?
For a long time usability research in the literature focused on colours, icons, task flows and scrolling behaviour. Less room was given to the text itself. Yet the user decides on the strength of the words they see on the screen.
Studies that examine this gap recommend using usability style guides at the development stage (Bowker 2015). A guide offers a shared benchmark both to the team writing the source text and to the team carrying out the translation.
One of the findings is surprising. Content written to both usability and translatability principles can lower the experience of the reader in the source language somewhat, while raising the experience of the reader in the target language. The interests of the two audiences, in other words, do not always point the same way.
This shows that writing the source text is a localisation decision. The texts of a product that will be published in several languages have to be written differently from those of a monolingual one. Taking the decision at the development stage is cheaper than trying to make up for it at the translation stage.
In practice this means that the team writing the source text and the translation team share the same guide. When headings such as sentence length, use of the passive, abbreviation policy and term selection are gathered in a single document, both sides work to the same benchmark. A short document gets read; a long one stays on the shelf.
How does machine translation affect usability?
One study that measures the question directly compared three versions of a word processor interface: the original English version, a raw machine translation version and a version localised entirely by human hand (Guerberof-Arenas et al. 2021).
The results showed that the mode of translation is decisive for usability. In the raw machine translation version, task completion times lengthened and the satisfaction users reported fell. No marked difference was seen in the number of tasks completed.
This finding should be read in two directions. Machine translation does not stop the user altogether, but it slows their work and weakens their experience. In text used every day, such as an interface, that slowdown accumulates over time, and differences that look small when measured one by one become pronounced in the aggregate.
A similar result holds for post-editing. Where post-editing is carried out, the difference between source and target readers drops to a negligible level. What is decisive, then, is not whether the machine is used but whether the output has been through human hands.
The Türkiye context
In Türkiye the English versions of corporate websites are usually treated as a one-off translation project. The site goes live, and the updates later made on the Turkish side are never carried across to the English side. The result is a difference between the two versions that is not only linguistic but functional.
Habits within Turkish itself also affect the measures. Users can recognise some English terms faster than their Turkish equivalents. Insisting on a purely Turkish rendering in such cases can lower learnability; the decision should be made by measuring users, not by linguistic preference.
The third point is measurement culture. In local projects a usability trial rarely appears as a budget line. Yet even a short task trial with five users can make the most common sticking points visible before the site goes live.
Where public services are delivered online, the matter is not only commercial. A citizen who cannot complete a transaction puts a load on the call centre and on physical branches. Simplifying the interface text can then turn into a measurable saving on both the user side and the institution side.
What does this mean for you?
When you buy translation, do not build your acceptance criteria on linguistic accuracy alone. Whether the product gets the user to the goal is also a measurable heading, and it is usually the more meaningful one for you.
The way to do this is not to prepare long reports but to put a few concrete steps in order. The list below can be completed in most projects with a week of preparation, and it can be reused for the languages that follow.
- Write down the three most critical user tasks: registration, purchase, support request and the like.
- Gather all the interface text that appears in those tasks in a separate file.
- Draw up a term list with a single rendering for each frequently used action.
- Review error and warning messages separately, and make sure they contain a suggested remedy.
- Before going live, have a few users perform these three tasks and note the times and the sticking points.
This work does not call for a large budget; what it calls for is getting the order right. For an assessment suited to your scope you can use the quote form, and you can reach the other related articles from the knowledge centre.
Frequently asked questions
How is usability defined?
According to the international definition, it is the extent to which a system or product can be used by specified users to achieve specified goals with effectiveness, efficiency and satisfaction. The definition always refers to a specified context of use.
What are the five measures of usability?
Learnability, efficiency, memorability, satisfaction and errors. The first four measure how comfortably the user learns and uses the product, while the last measures error-making and recovery from error.
Does error here mean translation error?
No. Error here is the wrong action the user performs while interacting with the product. Defects arising from development or translation are a separate heading, although poor text can increase user errors.
Does machine translation lower usability?
In one study, task completion times lengthened and the satisfaction users reported fell in an interface presented in raw machine translation. No marked difference was seen in the number of tasks completed.
Sources
- Jiménez-Crespo, M. A. (2024). Localization in Translation. Routledge, pp. 105–109.
- ISO 9241-11:2018. Ergonomics of human-system interaction: Usability.
- Guerberof-Arenas, A., Moorkens, J. and O’Brien, S. (2021). The impact of translation modality on user experience. Machine Translation, 35, pp. 205–237.