How can you supercharge your project managers with XTRF and empower your translators with machine translation? Nansija Lībiete (Head of Localization) and Rubén Martínez (Client Success Manager) from Tilde, a leading European language technology company, joined us at XTRF Summit #online 2020 to share their experiences. This article is based on their presentation.
Supercharging our project managers with XTRF
Project managers are a vital part of any translation or localization project. Given this, providing them with the tools to do their job as efficiently and effectively as possible seems like a no-brainer. And yet, across industries and sectors, project managers still face the same key challenges:
- Huge workload
- Difficulty assigning projects to providers
- Difficulty managing projects within the allocated time
So how can we help PMs overcome these challenges? How can we help them feel like they lead and manage processes, rather than the processes managing them? And how can we best use their precious time? Because at the end of the day, PMs are overqualified for completing simple manual tasks. It’s best to leave the system to do this administrative work and let the project managers bring real added value elsewhere.
I like the saying “Project Management is not what happens to us but rather how we make it happen.” But it’s not always obvious how we can make this ideal a reality. Here at Tilde, we use XTRF to help us. I’m going to focus on three core functionalities here: customization, analytics, and automation.
PMs rely on the information. The more helpful information you can give them, the more efficiently they can manage a project. So we use virtual columns, custom fields, and views to easily provide them with key information about providers, jobs, and tasks. Here are a few examples:
- (reference slide #6, #7) Automatic vs manual task and rate assignment. This shows which tasks and rates were assigned manually and which were added automatically. For those added manually, we can investigate why automation wasn’t used and fix any problems there might be. Then, the next time the PM works on a similar project, or with the same language pair, they can use the information added automatically. So it saves them time.
- (reference slide #8) Open and unassigned jobs view. This allows PMs to keep tabs on jobs that haven’t yet been accepted by providers. The view also includes the project deadline and word count so you can understand how urgent it is. Vendor managers and supervisors also find this view helpful.
- (reference slide #9, #10) Provider workload. Here we’ve added our own column which shows vendor workload as a percentage. We can also see more detailed information, such as the number of jobs each vendor has approved, rejected, or hasn’t answered, as well as any planned vacations.
- (reference slide #11) Job offer statistics. This is a really useful view that shows how many jobs are approved, rejected, or unanswered. For example, if we see we have a large number of rejections, this might prompt us to think about how we can improve our working relationship with our providers to increase the acceptance rate.
We can turn each of these views into reports and then use this information for various purposes. For example:
- (reference slide #12) Availability status. We can use the job offer statistics information as a basis for vendor managers to understand specific issues and needs. For example, they might see that there are a lot of tasks in specific language pairs, but also a lot of rejections. So they’ll want to delve deeper into why that is – is it related to specific clients? Particular tools? Or subject areas? This can help improve our working relationships with our providers and allow us to spot issues and needs.
- (reference slide #13) Timely deliveries. This helps PMs make sure they choose the best linguists by understanding linguist reliability. It shows how often different linguists deliver jobs on time. For instance, are there one-off delays, or recurring patterns? Maybe particular tasks are delivered late, or perhaps there are systematic issues.
XTRF offers a lot of options to automate certain manual tasks, freeing up our PMs’ time. Some examples of the functionalities we use are:
- (reference slide #14) Periodic jobs. These are different activities that are automated and just happen in the background. They include things like daily, weekly, and monthly reports of specific key metrics. We also automate the addition of categories.
- (reference slide #14) Workflows. We have different types of workflows suited to different types of projects. This helps us save time as we don’t have to establish this each time. One such example is for Memsource integration with MT, as we discuss below.
Empowering our translators with a machine translation
In the industry, the perception of MT has changed and evolved significantly over the past decade. It’s now seen as one of the most important productivity enhancements for human translators.
It’s also been interesting to observe the trends, expectations, and adoption of MT, from statistical MT through to neural machine translation and beyond. Today, MT is better than ever. And it’s being adopted by increasing numbers of companies. According to a 2020 Slator report, almost 7 out of 10 LSPs have either:
- Fully integrated MT, or
- Are midway through implementation
But one MT engine is not like another. Different engines suit different situations. In-house at Tilde, we now have almost 50 different MT engines. Let’s run through the different types of MT.
- Trained with large amounts of data from different industries, or specific domains
- Ready to use
- Adapted to a specific brand and domain
- To a certain extent, adapted to terminology, language, and style
- High accuracy and quality
- Retraining enhances the system qualities and capabilities
- Learns from translators’ post-edits in real-time
- Stylistic and lexical adaptations
- Terminologically more accurate
- Adaptation to the domain, customers, or even project
Future developments in MT
Integration is the name of the game. MT is now being requested and integrated into many different technologies and scenarios. For example, there’s an increasing demand for integration with websites, collaboration tools like SharePoint, and other systems like Microsoft Office packages. It will be interesting to see how this develops in the coming years.
Head of Localization Department
Client Success Manager