Some Notes From Our Journey Towards Project Management Automation

The Challenge

Technica Translations is a relatively young MLV that has spent its first three years concentrating almost exclusively on subcontracting translation and localization into the languages of the Baltic States (Estonian, Latvian, Lithuanian) for LSPs of various sizes all over the industry. In 2020, we served 80 customers, pretty much all of them on a daily, weekly or at least monthly basis. In 2020, we handled ca 7000 projects, a number that we are projected to easily surpass in 2021. The sheer number of projects seems to be growing exponentially, so here’s a little peek at how we’ve been trying to tackle this problem. I hope sharing this will come in handy for some of our industry peers and perhaps start a few useful conversations.

So, in the end of 2019 we found ourselves in a situation that many translation companies know only too well – while the language services industry is slowly consolidating businesswise, the actual work processes are more fragmented than ever. Our PMs often found themselves looking at 30 new projects in the inbox, the customer impatiently waiting for confirmation, while just processing the projects and entering them into our TMS is bound to take hours. This was clearly not sustainable.

First Steps

We use XTRF as our TMS. It’s currently the best fit for our needs, but like everything in this imperfect world, it has its improvement areas. Our first task was to identify the easiest changes in our workflow that would speed up the processes. Talking to our PMs, one of the things that kept coming up was the simple task of getting the basic information and files from the customer’s system into our own. This is a very repetitive, routine task that adds little value but cannot be skipped and will consume huge amounts of valuable resources like time and the nerves of PMs. We decided there had to be a way to work around this, started looking for solutions and found that there is not a lot out there.

We thought about creating web- or mailhooks to integrate our mailboxes into the XTRF workflow. There is usually a notification from the customer for every single step, which we sort into a nice little pile in our Outlook inbox (and never open again). So we thought these could be used to automate the information flow, e.g. by creating a script that would pull information from the customer’s system into our own via a web interface. It would be possible in theory, but not feasible or cost-effective for a company like ours – as our customers’ processes and systems vary considerably, we would have to create an unreasonably large number of custom solutions which will require time and money and will still break with every simple software update or change of process. Not a solution.

We considered several similar fixes, were not happy with any of them, and then decided to try out the BeLazy platform, which was still quite new at that time. In theory, it seemed to be addressing the exact problems we were trying to solve.

Implementation

Our initial, possibly slightly naive vision was to just add all of our customers to BeLazy and then sit back and relax while watching all the magic happen. Well, it is not quite that simple.

As it turned out, there were numerous issues that we had not foreseen. For instance, even if we are using the same TMS as our customer, there is no reason to expect that the workflows have been set up similarly. This brought up a host of problems that had to be solved, some related to BeLazy, some to XTRF, some to our own in-house processes and/or people.

As an example, we had to map the customer’s CAT grids to ours so that the system would know which fields will have to be populated with which information. So we had to adjust our XTRF workflows and customer/price profiles considerably.

Another problem arose with image files that some customers include in their instruction sets. XTRF would not add these pictures to our project instructions, but rather display long and messy lines of code. XTRF is aware of this issue and they are working on it, but for now, we had to ask our customers to include the images as separate files instead of including them in the instructions themselves.

We also spent quite a lot of time trying to get the receivables from the customer’s TMS to automatically translate to our system. We couldn’t figure out if this was an issue of mapping the connection properly or if there was something we had to adjust in our own TMS’s services, price lists etc. There was quite a bit of trial-and-error-back-to-square-one before we could get it working.

These are examples of the issues we had to tackle, but honestly in retrospect, all of this was to be expected. We worked in tight cooperation with the BeLazy team solving them one by one.

In addition to the systems and process specifications, there was also the not inconsiderable task of onboarding our in-house project management team. Language industry PMs are busy people with tight work schedules, and they understandably have little extra time to try out new fascinating gadgets, just to see if they might work this time. There was a bit of ‘selling’ and persuasion that had to be done before the PMs started to accept that the things we were trying to implement were actual solutions to their pressing problems rather than just another weird and useless wild goose chase. By now, however, I think I can say safely that at least some of our PMs would mutiny if I told them that hey, trial over, let’s get back to manual processes.

Results

For our current connections we have managed to get the average project importing process down to one minute. This means that a PM needs one minute to get all the instructions, financials, files, deadlines and other data from the customer’s TMS into our XTRF. This can still be improved of course, but can be considered a success already – when you have 30 tasks waiting in your inbox in the morning and 30 minutes later they are all confirmed and imported to your system, this is already a huge improvement from what we faced a year ago. We are clearly moving towards delegating repetitive tasks to the robots and our PMs can dedicate more time to the more humane and value-adding tasks of communicating with the linguists and the customers.

What We’ve Learned

There are quite a few lessons on various levels we as a team have learned from this process, and I’m sure there will be plenty more. Here’s just a few of them:

- There is no such thing as a standard project, ergo no such thing as a standard one-fix-for-everything solution. But there are useful tools that help smooth things out considerably.

- Every customer is different, every project is different, every TMS and project management tool is different. For example, there are XTRF to XTRF connections, Plunet to XTRF connections, there are huge LSP portals like those of Lionbridge, Translated or TransPerfect. The number of possible combinations is vast, not to mention variations between individual projects. This means that it is naive to hope for a universal savior that you can just switch on and see all of your problems disappear. You have to look at the entire lifecycle of each project separately in various aspects – from the customer’s TMS to an integration platform to your own systems and processes.

- You need a clear set of goals you want to achieve. Communicate these to your partners and (if possible at all) customers. Most of them are not huge faceless monoliths but partners open to discussion once they understand that you have the same goal – to make project administration as smooth and quick as possible.

- When you start working with an automation partner like BeLazy, make sure that you are constantly on the same page, and this includes not just your in-house developers but also the PMs. The PMs are the ones who will be using the end product of your endeavors.

- Focus on two or three customers and connections at once. Get them set up as well as possible, learn from the mistakes and move on to see how others can be integrated.

- An integration platform like BeLazy is only a part of the solution. You really need to look at the processes as a whole, from the first confirmation in the customer’s system to the final handback. You will always find more things you can automate and more clicks you can eliminate.

The process of moving towards automation can be called a mindset. The whole team has to buy into it, everyone can come up with ideas for improvement. An integration platform is just one step, full automation of everything can never be ‘completed’.


For further reading, here's BeLazy's view of our joint efforts in 2020