On XTRF and Scaling a Translation Management System with Andrzej Nedoma

Translation Management System XTRF - Andrzej Nedoma on SlatorPod

Andrzej Nedoma, Co-founder and CEO of XTRF, joins SlatorPod to discuss the journey of the translation management system (TMS), from stepping out on his own to teaming up with private equity.

Andrzej begins with XTRF’s early days as a spin-off of family-owned LSP Lidolang and building the company with his Co-founder, Dominik Radziszowski. He talks about the rationale behind pivoting from licenses to SaaS in 2014 and the funding decisions with Experior Venture Fund, whose shares they bought back at the end of 2020.

The XTRF CEO unpacks the process behind investment firm K1 acquiring a majority stake in the company and their relationship with sister company XTM going forward. He talks about the similarities and differences between LSP and enterprise clients, and how they deploy their systems to fit users’ needs.

Andrej shares the key milestones and shifts in the TMS landscape over the past decade and XTRF’s most important innovations in the last five years, including building their Business Barometer. He rounds off with XTRF’s roadmap over the midterm and the impact of machine translation on TMS in the broader industry.

First up, Florian and Esther discuss the language industry news of the week, with two UK-based life sciences companies teaming up, as LSP Conversis acquires local rival Zebra Translations

Esther reviews Slator’s latest Language Industry Job Index, which climbed by 3.8 points in October, after a 10-point upsurge in September. Florian touches on Slator’s Language Industry Buyer Tracker, which includes a new Localization Producer for Games at Netflix, a move that possibly indicates the company’s plans to expand into the gaming market.

It’s been a busy week for Netflix as the streaming giant got called out for botched Korean into English subtitle translations for the TV drama Squid Game, which is on track to become the streaming giant’s most-watched show of all time — indicating the rising demand for English-dubbed content.

SlatorCon Remote June 2024 | $ 180

SlatorCon Remote June 2024 | $ 180

A rich online conference which brings together our research and network of industry leaders.

Buy Tickets

Register Now


Florian: Let us talk a bit about XTRF in a nutshell. Tell us a bit more about the origin story, founders, trajectories, key clients, just the extended elevator introduction.

Andrzej: The story behind XTRF starts 20 years ago when I finished my studies and I joined my parents in their home-based translation company which they wanted to grow. I was invited to help them grow this company, which was great and we happened to grow from the one-man show of my father, providing old translation services to one of the most important translation companies in Central and Eastern Europe. This was Lidolang Technical Translations back then and as we were developing the translation company, we saw how much we needed technology to automate our operations because project management was not super-efficient. Whether it is about preparing files or calling translators to see who is available and then storing every data related to pricing these clients, pricing these vendors, and then executing on the project. 

A lot of work typically happens in a translation company to execute on a project and we need to do something about this. Back then we did not find the tool which was suitable for what we planned so we decided there was an opportunity on the market to build a tool to help translation companies with this stuff. Since we were part of different international organizations back then, I was talking about this idea with lots of CEOs and managing directors of different companies and they were all super excited. I said, there is a niche, there is potential, we should go for it, and this is how our idea of creating the technology spinoff from the service company was born and we decided to do it. We were bootstrapping in the early days and developing this software by covering the costs from our own pockets and our own money. 

In 2010 we decided the company is becoming serious and we have more and more developers, so we decided to split the companies. This is when Lidolang kept focusing on delivering the translation services and XTRF was created with purely technology and product. This is how we started to grow, serving translation companies and providing help with the tool that we were offering. We have seen the potential also in corporate clients, basically corporate translation departments that behave like translation companies internally within big organizations so we knew this is the market we could go after. We realized this much later but today we have those two groups of clients that we are serving and in 2014, we decided to raise venture capital and expand our organization. This was the moment where we also wanted to change the model from selling licenses to software as a service so renting licenses, which changed the financials and dynamics in the company. In December last year, we bought back the investors package and we became fully independent and recently we have announced the majority investment from K1 who joined XTRF. 

Florian: There is so much engineering talent in Poland, was that a factor in influencing your decision to do a standalone software company or did this have no influence at all?

Andrzej: It was important and useful, but this was not a factor in the decision to start the business. We had to hire developers but it was not Kraków as a location that invited us to build a tech startup. The co-founder, Dominik, has a PhD in IT or information science so he was the one responsible for building the team and he is the brain behind all technology concepts in XTRF. Him being a PhD and also working at one of Kraków’s universities gave us access initially to talents that we could source and bring to our company. This was how we started.

Esther: On a personal level, what was it like for you to step out on your own when you made the decision to spin XTRF off from the family LSP?

Andrzej: Yeah, it was exciting building the translation company with my parents and then building the software company. It is an exciting evolution. I was involved a lot in the growth of Lidolang, so I was still helping them with strategy and helping them organize things, but also building my own product, which was invented by me and Dominik and luckily co-sponsored by my father who was the initial investor in XTRF. I was lucky to have a lot of trust from my father so it helped and worked very well having him as a partner as he was one of the shareholders in XTRF because we were driving our revenues from the translation company to build a technology startup. It meant that we are all working together, but it is true that this was the first company where I was the CEO  from the beginning and was behind the strategy and go-to-market and the product idea.

Florian: Tell us a bit more about the decision to go from licenses to SaaS. It was around 2014 you went full-on SaaS. How was that decision? How was the background? How hard was it? 

Andrzej: It is very hard not to do it because when you sell licenses, you have to chase clients to generate revenue so that you can cover the costs all the time but once you have sold, you have only a tiny support fee that comes back to you every month or every year. Switching to rental means that you build a stable recurring revenue, which requires effort and time to build this recurring revenue to the amount which covers your cost. Once you are there, your financial stability is much higher than in the other model. Of course, there is a gap in between because when you stop selling and you start renting, suddenly you have much less cash because renting is cheaper than selling. You need cash to fill this gap and this is why we decided to cooperate with VC because we needed this money to execute the plan. 

This was the story behind our pitch to venture capital and SaaS was already becoming well-known in the world, maybe not in translation space, but in lots of spaces and lots of startups, this was becoming the most popular model so naturally, we want to go there. It is not like we got there straight away, because first of all, we had clients who got their licenses before we switched. Then there were customers who were still reluctant to switch to this type of business model and then we were phasing out the old one and bringing the new SaaS model more and more to new clients as we were acquiring them. It is not like everyone today is on the SaaS model with XTRF, but all new clients are. 

Esther: It is interesting there that you link together the VC funding and the pivot to SaaS. Was the funding very much contingent on that move to SaaS or vice versa? How did it work? 

Andrzej: That is what you need money for. Of course, we could spend all the money on marketing and try to sell more but as I said, that is great the months that you sell but next month you still start with very low income and you have to generate the next income to cover the cost next month, next quarter, next year and it keeps going that way. Our strategy was to grow in the new business model and we knew it needed money. Otherwise, we would not be able to cover our costs because, as I said, without having a huge income as a result of selling licenses, we would have no money to continue our operations and grow the business. As in every startup, you have to cover two sides of operations, one is investing in getting customers and sales, and the other is keeping the development of the product going because every client comes with new ideas. We also had our ideas about where we want to go, but the more clients we have, the more ideas we get about what they expect the system could do for them and how we could develop that. There is always a huge appetite for investing more in the tool so you need money on both sides of this. Keeping balance is not easy and changing the business strategy or business model, at the same time, is even less easy, so that is why we needed a VC to help us with this. Without VC we would not be able to finance that. 

Florian: Over the past 5 to 7 years, where did you see the most effective channels for doing marketing and sales? 

Andrzej: I have the feeling there is no general rule about what works in marketing. It changes over time, each year is different. Going to conferences was great, we were visible and we could meet a lot of people, but then we realized that there is always more or less the same group of people that are attending conferences. Then we tried more online marketing and outreaching to clients. Then it was social media and content marketing, which probably works for us the best in the recent two, three years and recently proactive contacting of clients. Combining content marketing with marketing automation, then engaging those found leads into more and more advanced discussions. This is what is working best today. 

Florian: Someone who might give some funding for all that sales and marketing is K1, which is a California based PE firm. Are they growth equity or more mature private equity?

Andrzej: That is the beauty of this partner. They are private equity, investing in companies with high growth potential and software as a service that delivers mission-critical software to their clients. We perfectly fit into this category. It means for XTRF that we get a partner who is interested in growth, which is great news for the team and for clients because they want to grow with XTRF. Not to mention this is also great news for our biggest clients who are always afraid of what would happen if we got bought by a big LSP, which is not an improbable scenario, but while getting this investment from K1, this is off the table. Having a financial partner in private equity and investing in technology, gives much more security to all our clients that we become vendor-independent. We are not offering translation services. We are not competing with them in any way so this is great news.

At the same time having a partner who invests only in this type of business means that the experience that we got access to is stored across the whole portfolio of companies that they have, so this access to knowledge is huge. I can see everywhere that we have similar types of questions and issues that we face growing the business. Here we have access to over 50 companies who do more or less the same type of business that we do. Of course, in different spaces, in different industries, but being SaaS, mission-critical, and growing more or less in the same business model. There are a lot of similarities and we can leverage this knowledge.

Florian: When we saw the announcement with K1, we thought, is not that the same fund that also invested in XTM? Can you just tell us a bit more? Is there any overlap? Was that part of their decision to go also with you guys? How did it unfold?

Andrzej: The fact that we are sister companies creates a lot of potential and benefits. Of course, we run as independent businesses. It is like living under the same roof because K1 is our partner, but XTM was always much more into corporate clients, so we felt that to enter this market, we needed the CAT tool. We needed a CAT tool partner and we did not plan to develop a CAT tool component within XTRF so now we see the potential of addressing this market together. In terms of the LSP market, of course, this used to be the core market for XTRF and we have lots of connectors with different CAT tools there. This will not change the market, especially LSPs need to work with lots of CAT tools so we need to maintain all the connectors so it is hard to say that this partnership changes anything from this perspective. We will be improving our connector with XTRF because being sister companies, we can align our production cycles and roadmaps so that it provides even better value for the market. 

Esther: You mentioned a bit about the LSP customer base and then the enterprise clients. What similarities and differences do you identify between those two customer profiles? 

Andrzej: They are rather different than similar. We provide them with the same tool because we have only one core XTRF, but they look at this tool from different perspectives. The emphasis on different features and functionalities is different. LSP companies serve hundreds of different clients so they have hundreds of different workflows. They work with lots of vendors, thousands of them in the vendor pool that they maintain, so they have lots of unique types of projects. While corporate clients serve their internal needs so they probably have a lower number of different workloads that they execute. For example, the usage of our customer portal might be different because for the corporate clients, what we call a customer portal, they call a procurement portal because, for them, this is the tool for internal users to centralize all the requests on one content management office or localization center. You can see that they use the same elements available in XTRF for different goals. They would also typically work with fewer vendors. They tend to have just a few LSPs that they down source the projects too, so there are differences. We managed to maintain one core system that can be adapted to the two types of clients that we have. That is what we handled for so many years, how to best create the tool which can be quite freely and openly developed or customized due to different needs and user scenarios that we see across all the customers. Even between similar groups like LSPs, they all have their particular ideas and they might be treated differently from each other so the tool needs to be open enough, flexible enough to be able to accommodate those differences.

LocJobs.com I Recruit Talent. Find Jobs

LocJobs is the new language industry talent hub, where candidates connect to new opportunities and employers find the most qualified professionals in the translation and localization industry.

LocJobs.com I Recruit Talent. Find Jobs

Florian: What are some of the key differences for deploying the system? For example, doing continuous training, onboarding new people, working on bespoke connectors?

Andrzej: The smaller the company, the fewer customizations they want to have, or they can afford to have. They work in a more or less standard way, and the bigger the size of the client as the company grows, the more sophisticated the needs and the more particular their end-client’s needs are that they have to accommodate. As their technology skills grow or as they build their technology departments, they have more ideas about what additionally can be automated, and what can be built on top of XTRF. Their requirements related to some particular scripts or macros or other applications that can be developed on top of XTRF are even more sophisticated. That is what we help them do as well.

Florian: Do you have a strict hold the line policy in terms of not having feature overkill and trying to accommodate everybody’s niche features? Does the platform become super heavy? 

Andrzej: We maintain one core system so there are no different versions of XTRF and each time when the customer asks for something special, we try to understand whether this is unique for the customer, or this is quite generic for the market. When this is generic for the market, it is a question of how can we implement this within the core system? We are very selective about what we decide to implement so that we do not add 1000 buttons on one page, which is a risk there. This risk can be managed by understanding that lots of these requests are unique to clients, and if the platform is flexible like XTRF is, we can build the macros, scripts, small applications or some custom tools that can be attached through the API. It means that we develop a little piece of software just for this customer, so it is not visible from the main user interface, but it delivers a certain value or feature to this particular client. In that sense, we maintain one core system, but we can have lots of different applications or customizations that we develop for clients individually but as long as they use the same API, their maintenance is not all that difficult.

Esther: What have you observed over the past decade operating with XTRF? What have you seen as some of the key milestones and shifts in the TMS landscape generally? 

Andrzej: First is that nobody really knows what a TMS is because we use this term for so many different tools like XTRF or business-oriented tools or CAT tools. Now, these words merge, and we can see a lot of functionalities overlapping. At the same time, when we started with XTRF, people were reluctant to automate everything. They were rather using this to keep data in order and keep track of everything. It was like a recording tool that was recording all the business projects, events, and processes. As time goes by, people are more and more open to automation and willing to automate, because the type of projects on the translation market changed. 

10 or 20 years ago we would see many more projects that included huge manuals to be translated into several languages and we had a few weeks to do that. Now we do not see those types of projects so much because there are lots of small bits of texts that come from content management systems that are translated into many more languages, the deadline is within hours or days, and there are lots of steps involved in the process. We need to execute lots of operations, which would not be possible without automating them. It would cost too much for the tiny value of those small projects that come frequently every day or several times a week. That type of project and the usage of CMS’ imposed a different usage of business management systems or project automation systems like XTRF. 

Esther: In the past four or five years, what do you think are some of the most important two or three innovations from XTRF?

Andrzej: First would be the automations made available to customers. Second, is the development of what we called smart projects, a new interface for managing projects. It is much nicer in terms of the user interface, usability, and how we present or give access to different features. Not everything is on the screen immediately so we show the functions that are important or needed by the user in a specific situation, specific moments, so then we do not charge the screen with too many options at the same time. This is something that wins a lot of clients for XTRF, the smart projects, beautiful interface, very responsive, very attractive. Then the cloud, so XTRF is a single-tenant application, which means every instance of XTRF is separated from the other instances, so there is not any risk or flow of data between the instances. At the same time, there are new methods of developing tools when you base it on the cloud and everything which is available as tools within the cloud, which can accelerate the deployment of new applications by far. 

If you look at what happened during the start of the pandemic, we saw lots of our clients panicking about what is going on with their money, the clients, the projects, and they all had one or two questions in mind. Which clients of theirs go down? They would need to understand the situation, assign customer success teams to try to find the solution and help the clients somehow. The other question was which of their projects or clients go up and despite the pandemic, keep ordering more because they are from pharma industries or similar? They could generate upsell if they concentrate the salesforce there. They could somehow make up for the loss that they are experiencing on the other side. We said we need to provide very quickly a tool that would present dynamic trends to our customers about their businesses, their customers, and what is happening day-to-day. 

We developed the XTRF Business Barometer, which is a reporting tool based on trends of different business parameters or key performance indicators, which could be lifesaving for some of our clients. Giving this information quickly to them was important and this was developed in the cloud. We have seen how within weeks we can provide new functionalities and deliver new value to the customers and we focus on developing more and more of these cloud solutions, which develop much lighter in deployment and can deliver value faster. I think that is the way to go, so we try to position new components in the cloud. 

Florian: We spoke about the SaaS business model, which is super attractive because it has recurring revenues. Was it as stable as you would have expected during the early part of the pandemic? 

Andrzej: When you are in software as a service model, you have a few KPIs that you monitor and one of them is churn, which means the clients from your portfolio that leave you. This is the main factor that we are monitoring, but we could see nearly zero churn and a little bit of down-sell, which means customers had to let a few people off so they were reducing the number of licenses in use, so at a little bit of down-sell could be seen, but very limited. We have already seen it go back to normal after a year. I must say we are in good shape and we did not have to panic at all. This was the moment where we felt the financial stability of the company, of employment, and the business in general.

Esther: Have you seen any impact from machine translation on TMS development generally? How are those two interacting now? 

Andrzej: Yes, because machine translation touches the content, it helps the translation process so typically this is plugged through the CAT tool. In XTRF, the tool that manages the whole project part of this workflow is related to real typing of the translation or editing or quality check. This is outsourced or done in cooperation with third-party technology to us, which is a CAT tool. For this CAT tool, typically you would plug the machine translation because rather than doing human translation, you can implement the machine translation step and then have the post machine translation editing step done by humans. From our perspective, being more of a logistics oriented company, it does not change much. Of course, the names of the steps change. We can envisage processes that just require machine translation or just short editing of this machine translation output. We could see the connections between TMS and machine translation directly and deliver it to the customer, or plugging a simple editor just to edit the output from the machine translation and without benefiting from the whole translation memory system. 

Florian: Now that you have access to that big network of expertise and capital from K1, what is your roadmap for the next three to five years? Where do you see the broader industry going?

Andrzej: I can see that the types of projects that require human translation and machine translation are pretty different. There is lots of regulated content, in life science, in legal, in finance, in banking and we surely require a lot of attention to detail there and we can see that the role of humans there will remain important and there is a lot of focus for different LSPs on these areas. While technology or manuals or this type of technical documentation is probably more ready for engaging machine translation because surely machine translation is what impacts this industry the most. Of course, it is hard to say that it will leave people without work because as we all know, only a small fraction of content produced worldwide is being translated because there are not enough people and not have enough money to finance that so machine translation will help fill this gap, but surely we will see the shifts in which areas people are engaged and which areas are more ready to engage machine translation. Then AI, NLP and data science or data-related services. This will be booming. You cannot stop it. 

As for the next three to five years, that is too long of a period to make a justified prediction. Mark Twain once said predicting the future is difficult because it is related to the future. It is hard to say what will be there, but I am sure usage of machine translation will be more and more natural. Also the standards and acceleration of all processes towards the continuous localization process, and hiding the technology from the user perspective because technology is there to execute the process in the background. We would rather be setting up workflows that will be applied later so that everything can happen automatically rather than building a workflow as a response to client requests for a particular project because everything would happen too slow in such a scenario. Preparing technology first and then executing on the flow of content so that it is processed automatically.