Introduction to Quality Management Systems (QMS)
Quality means different things to different people. To illustrate that, let’s start this lesson with a poll. Head over to Poll Everywhere and respond to the discussion question.
Discussion Question What does quality mean to you? Answer on Poll Everywhere: pollev.com/alainabrandt831
To make the subjective idea of quality objective, we can start by having discussions about goals and needs, define requirements in specifications, produce work in keeping with specifications, and then measure our outcomes against specifications. If what we produced does not correspond to our specifications, or if the work corresponds to inaccurate specifications, we don’t have quality work. In this context, quality management systems (QMS) are about refining specification setting and production methods so that workflows have good quality outcomes. Here good quality outcomes are consistent and the meet -or surpass- stakeholder needs.
From Translation Quality to Quality Management Systems
The following section examines quality management principles through the specific lens of translation and localization work. While the examples focus on translation quality—a topic directly relevant to your future careers—the underlying concepts apply across industries and business contexts. As you read, pay attention to how quality management works in translation workflows, but also consider how these same principles might function in manufacturing, software development, healthcare, or any other field where systematic quality control matters.
The goal is twofold: to understand quality standards in your chosen profession and to recognize the universal principles that make quality management systems effective in any context.
This chapter began with a discussion about the fact that every role – no matter the industry or level – entails project management. While the assessment of translation quality often focuses upon the work performed by the translator, translators are just one part of larger production processes, and mishandling the product at any stage compromises the quality of that product. If production is underquoted, for example, that has a downstream impact on quality. That said, all stages in localization workflows are a part of quality management. Then again, certain steps within localization workflows are quality management proper, such as the editing and proofreading steps in a standard TEP workflow, and any in-context reviews that take place as translated content is imported into the localized product environment.
In the ISO 9000 family of standards on quality management systems (QMS), quality is defined as the “degree to which a set of inherent characteristics of an object fulfills requirements” (ISO 9000 QMS - Fundamentals and vocabulary, Section 3.6.2). An overall QMS consists of many parts, and quality management is the combination of quality planning, quality control and quality audits (among other activities), which when combined can produce continual quality improvement. These activities embody the Plan-Do-Check-Act cycles recommended in ISO 9001 - QMS Requirements (Section 0.3.1), in which production is planned, production is carried out according to the plan, the product that is the result of production is checked against the plan and any opportunities or gaps within the plan and how the plan was carried out are acted upon as the cycle begins again. Overall, quality managers should be thought of as process managers’ best source of information about whether production as configured is producing intended results.
Adapted from Alaina Brandt, “Translation and Localization Project and Process Management” in the Handbook of the Language Industry, De Gruyter
Note: You may have access to the chapter “Translation and localization project and process managers” by Alaina Brandt if it has been shared with you privately in keeping with fair use provisions defined in U.S. copyright law. If you have this free access, please do not share this resource outside of our class.
Making Connections: From Translation Quality to QMS
Consider the following questions after reading about translation quality management:
- Universal Principles: The passage describes quality as “the degree to which a set of inherent characteristics of an object fulfills requirements.” How might this definition apply to products or services outside of translation? Choose an industry different from translation and explain how this definition would work in that context.
- Upstream Effects: The text notes that “if production is underquoted, that has a downstream impact on quality.” What does this tell us about how different stages in any production process affect final quality? Can you think of examples from other industries where problems early in the process compromise the final product?
- The Human Factor: Translation quality depends on multiple roles beyond just the translator (project managers, editors, proofreaders, technical specialists). Why do you think quality management systems emphasize the entire process rather than focusing only on the people who create the core product? How might this process-oriented approach change how we think about responsibility for quality?
- Plan-Do-Check-Act in Practice: The ISO 9001 standard recommends continuous Plan-Do-Check-Act cycles. Looking at the translation workflow described (planning, translation, editing, proofreading, in-context review), how do you see these cycles working? Where else might this cyclical approach to quality improvement be valuable?
- Your Future Role: As future translation, localization, and interpretation professionals, how does understanding systematic quality management influence how you think about your work? What responsibilities for quality might you have beyond producing accurate translations?
Correcting Quality One Non-Conforming Output at a Time
A non-conforming output, or a non-conforming product, is one that does not meet requirements, given the specifications and the developmental step at which point any non-conformities are identified. Non-conformities are essentially errors in the output, but not every output with errors is considered non-conforming. For example, in production workflows, we expect that a translation will be delivered with some errors. That is why editing is planned after translation so those errors can be rectified. A non-conforming output is therefore one that surpasses the error threshold for the given step.
Identifying Non-Conforming Outputs
Non-conforming outputs can be identified in a number of ways: by establishing what are the acceptable margins of error, and through error counts and time based metrics. In manufacturing environments, tolerance limits (the upper and lower boundaries of acceptable variation) are often established, and a workflow is considered successful if it only produces non-conforming output a very small percentage of the time. For machine learning tasks, often 95% accuracy is often considered to be a good achievement point where any gains in accuracy are not worth the computational costs. In translation and localization, often error counts and time based metrics are used as indicators of quality.
Within the error counting methodology based on systems such as Multidimensional Quality Metrics, errors are identified, categorized and counted, with critical errors being those that could cause physical injury, reputational damage, etc. Here, limits on the number of errors in each category that will be considered acceptable are tied to timelines and costs, where rush projects paid at low rates obviously realistically are likely to have more errors.
| Level | Number |
|---|---|
| Critical | Number |
| Major | Number |
| Minor | Number |
| Neutral | Number |
Time-based metrics are another way of determining if an output has surpassed the acceptable error threshold. Metrics are equations in which a unit of measure is multiplied by the time needed to process that unit. Metrics are used to determine how long a step should take, given the volume. The establishment of metrics requires time studies on processes, and an explicit definition of the level of quality to be produced.
| Step | Metric |
|---|---|
| Translation | 300 words/hour |
| Editing & Proofreading | 1500 words/hour |
| In-Context Review | 15000 words/hour |
| Final Review | 45000 words/hour |
Let’s say we have a metric of 1500 words per hour for the editing stage. Under this metric a project of 750 words should take a half hour to edit, and a project of 3000 words should take 2 hours to edit. If an editor needs more time than the time allotted by the metric to complete the editing stage, the work is identified as a non-conforming output, assuming that the editor is proficient at editing.
Following the examples given here, if an editor needs 1 hour to edit a project of 750 words, that equals a loss of one half hour. If an editor needs 4 hours to edit a project of 3000 words, the equals a loss of 2 hours. In both cases, the time beyond the amount of time allotted by the metric is the time lost.
Recuperating Financially from Non-Conforming Outputs
Measuring non-conforming outputs according to time is a simple way to tie a monetary value to time losses. If our organization’s hourly rate is $75 USD, we can next calculate our monetary losses as $35.70 in the first example and $150 in the second example. Once we have a dollar amount, we can alert a contracted translator if the root of the non-conformity was poor performance on their part that we will be deducting these monetary losses from the purchase order that they were to be paid, thereby covering our losses.
Any organization will likely want to recuperate the monetary losses resulting from reworking the output when an external contractor is responsible for the non-conformities. (To determine if the external contractor is responsible, the project manager or a quality manager will first need to do a root cause analysis. More on that in the next lesson.)
Here’s where the terms and conditions that govern the work are important. We covered the terms and conditions in Independent Contractor Agreements (ICAs) and Purchase Orders (POs) in Week 4: Legal Language in Business.
Contractor agrees to provide services in accordance to the Organization’s specifications, to complete all assignments previously accepted by Contractor and to have work reviewed by the Organization and/or an independent third party. If it is determined that Contractor has returned, in the Organization’s sole opinion, sub-standard, deficient, or incomplete work, the Organization has the right to withhold and/or reduce payment and to terminate the use of Contractors services.
Terms & Conditions from ICA or PO
If an external contractor is responsible, the monetary losses suffered due to the non-conforming output that they delivered can be recovered by reducing the payment that the organization agreed to pay for the contractor’s services. Reducing the payment can be safely done if the terms and conditions within the Independent Contractor Agreement (ICA) or the Purchase Order (PO) identified reduced payments as a consequence of non-conforming deliveries.
The contractor should be issued a new PO as quickly as possible in which the payment reduction is reflected, and the project manager should explain why the payment was reduced and the methodology used to determine the amount of the reduction.
We have a non-conforming output. Now what?
Start from Scratch, Rework, Use As Is
In discussing recuperating costs, we’re getting a bit ahead of ourselves. Before we can ever go on to recuperating costs, we need to know what to do when we’ve found a non-conforming output. When we suspect this, we have three choices to address the issue: start from scratch, rework, or uses as is. What does this decision making look like in production? Let’s consider an example from the world of translation.
Thinking about the team that will work together in translation and localization production, a project manager will oversee the project. A translator will do the translation. A quality reviewer will do the quality reviews. As soon as a quality reviewer suspects that the output they are working with is non-conforming, they should alert the project manager. This allows the project manager to decide what to do.
In many cases, the option of “rework” will be selected. In this case, the translator may be given a chance to rework their own output, or if the translator cannot, the rework may be assigned to another translator, or if the quality reviewer can make the necessary changes, the project manager may decide that the QR should do the rework and go over the time budget for their task.
In the case that “start from scratch” is selected, the PM will assign the work to a new translator, who will start the work all over again.
In the case of “use as is”, the PM or the account manager will explain the situation to the client and get approval over adjustments to the specifications to reflect that a lower level of quality will be delivered.
In any of these situations, if buffers have not been built into the timeline or if the buffers do not cover the extra time needed, the client will need to be alerted to delays in the delivery.
Initiating an Investigation into the Non-Conforming Output
After the production team has achieved the immediate objective of bringing the non-conforming output up to acceptable quality threshold in order to get the project back on track, each step impacted by the non-conforming output may be logged into a tracker of some kind in more mature production environments. This is done so that total losses (or gains!) can be tracked and to initiate an investigation into the non-conforming output by the quality manager.
Non-Conforming Output Tracker - Positive Feedback and Constructive Criticism
| Project No. | Date | Task | Time Allotted | Time Needed | Over/Under? | CA or PF* | Notes |
|---|---|---|---|---|---|---|---|
| 2025-100 | 2/29 | Proofreading | 3 hours | 5 hours | 2 hours over | CA | Punctuation rules from style guide were not followed. I had to fix all the punctuation. |
| 2025-103 | 3/1 | Format Review | 8 hours | 16 hours | 8 hours over | CA | There was font corruption everywhere. The DTP Specialist needed me to identify each instance and note the text that should be inserted where the tofu was. The DTP stage will be over time too, and a longer in-context review will be needed. |
| 2025-105 | 3/4 | Proofreading | 8 hours | 6 hours | 2 hours under | PF | The translator and the editor did a really nice job. I didn’t have to change a thing. |
*CA = Corrective Action; PF = Positive Feedback
Root Cause Analysis and Corrective and Preventative Action
The goal of investigations into non-conforming outputs are two-fold: we need to correct the non-conformity, and ideally, we’d like to prevent the same non-conformity from occurring again. To adequately correct and prevent issues, root cause analysis (RCA) must be performed to ensure we’re addressing the issue at its source. Methods for carrying out root cause analysis are addressed on the next page, and your assignment for this week will be to identify a non-conforming output, conduct RCA, and establish corrective and preventative action plans. These methodologies, when put into practice in production workflows, allow for continual quality improvement over time.
While the focus of this module is investigating outputs that do not meet acceptable quality thresholds, outputs that exceed acceptable quality thresholds need to be addressed as well. When outputs exceed expectations, it may be that the level of quality provided is too far above the rates being paid for the output. Or an excellent team combination or a better way of running a process may have been discovered, and that excellence is something you’d like to investigate and replicate!
📥 Download this Content
Find this file on our repo and download it.
🤖 GAI Study Prompts
Copy the downloaded content and try it with these prompts:
- “What are the key responsibilities of a quality manager versus a project manager in a production workflow?”
- “Compare quality management approaches in translation/localization to quality management in [industry of your choice].”
- “Create a scenario where I need to decide between ‘start from scratch,’ ‘rework,’ or ‘use as is’ for a non-conforming output, and walk me through the decision-making process.”
- “Quiz me on the difference between error-counting methods and time-based metrics for measuring quality.”
- “Help me understand how to calculate monetary losses from non-conforming outputs using time-based metrics. Give me practice problems.”
Next Activity: Root Cause Analaysis