One of the most important decisions you will need to make is to choose a translation platform that meets your current and future needs. All of these platforms do the same thing, which is to make translation work automated and manageable. The difference is mostly in how they do this and what other systems they are integrated with.
Most TMS implementations fail not because of the tools themselves, but because they are introduced before the underlying localization pipeline is properly designed. Even advanced and very well funded companies often start localization in a reactive and ad hoc manner.
What Is A TMS?
A translation management system is best thought of as a version of Github for translation. This isn’t a perfect analogy but it is easily understood by EPD decision makers. Key features:
- Act as a central repository for translations and related assets.
- Route translation work to different agents or service providers.
- Provide an editing and task management environment for translators and reviewers to work within.
- Provide integrations into other platforms. This is where there is considerable variability between vendors.
- Provide REST APIs that enable customers to build custom integrations.
All of the platforms mentioned support AI and machine translation as an option. The localization industry has been using these technologies for many years (LLMs are a direct descendant of neural MT technology which was introduced in 2017).
The TL;DR
This review focuses on cloud-native, API-first platforms used by technology companies; it does not cover agency-oriented or on-premise systems. Among the leading TMS vendors in this category are CrowdIn, Lilt, LocalizeJS, Lokalise, Phrase, Smartling and Transifex.
We’ve worked with all of these systems and each has its strengths and weaknesses.
CrowdIn – best for EPD led teams that are starting out with localization.
Lilt – best for enterprise and public sector entities that have large volumes of material to translate at scale.
LocalizeJS – best for dealing with legacy assets and web services that require extensive refactoring to support localization (can be used in parallel with other TMS solutions)
Lokalise – best for EPD led teams that are starting out with app localization
Phrase – best for growth stage companies that are committed to localization and global expansion over multiple years.
Smartling – best for growth stage and enterprise companies that are committed to localization and global expansion across user touch points.
Transifex – best for EPD led teams that are starting out with app localization, perhaps with plans to include other user touch points.
Choosing the “best” TMS is less about features and more about the integrations they provide to your product and engineering workflows.
This can all be a bit overwhelming because of the dependencies involved, so feel free to get in touch with brian@loctechpartners.com
Key Features & Differentiators
- Workflow automation – this enables you to put translation work on auto-pilot, where it is done by AIs, humans or both.
- Integrations (CMS, CI/CD) – a TMS that is poorly integrated with a CMS will become a bottleneck, especially as content volume and the. number of target languages increases.
- MT/AI support – all vendors provide this, and most make it flexible and route to the best agents for a given language and content type.
- Versioning – important for preserving a history of edits and the ability to roll back.
- Analytics – this gives you visibility into how vendors and service providers are performing and enables you to find and fix bottlenecks and quality issues.
CrowdIn
CrowdIn is a tech forward, developer focused TMS that offers AI first translation workflows along with a robust set of integrations with popular CMS, help center and other platforms. They also offer low/no cost plans that are adequate for building proof of concept projects.
Best for EPD led startups and growth companies that are just starting to experiment with localization.
Strengths
- Low/no cost self-service entry points : this is big for EPD led companies that want to build a proof of concept before committing to anything
- Robust integrations : CrowdIn offers integrations to a wide range of code repos, CMS platforms, and other systems
- REST API : they also offer well designed APIs so you can build custom integrations to other platforms
Weaknesses
- Self-service focus : great for EPD led teams, but marketing led efforts that require more support may find this frustrating.
- Limited US time zone support : they are headquartered in the Baltic region, which is great for Euro zone customers, but makes scheduling meetings in the Americas a stretch.
Cost
CrowdIn offers a free entry plan which supports up to 60,000 hosted words and one private project. Like other TMS platforms the subscription cost is based on the amount of content hosted or processed. Paid plans start at $50 per month with 2 private projects.
LocalizeJS
LocalizeJS is an interesting company. They provide a low code Javascript based tool for localizing web sites and web services that would otherwise be difficult or expensive to refactor for localization. This is a great tool to have on hand. For example, most companies I have worked with had static web content that lived in a repo somewhere. There isn’t a good way to localize these assets, but with LocalizeJS, this is a matter of pasting a JS snippet into page headers. As easy as adding a Google Analytics tracker.
Best for dealing with legacy assets and web services that require extensive refactoring to support localization.
Strengths
- Easy, like no-brainer easy, to add to legacy websites and services
- Great for prototyping localized services, even if you plan to do proper refactoring later on
- Developer friendly and relatively low cost of entry
Weaknesses
- Focused on this specific use case, not really a fully featured TMS so you’ll probably want another TMS for CMS, help center, etc which is OK
- In browser rendering limits search functionality, SEO visibility, etc, but that’s also OK for the intended use cases
Cost
LocalizeJS offers a 14 day free trial, and an entry level plan for $50/month, which also makes it easy to try without an upfront commitment. Since they sit behind a CDN, monthly costs scale with site traffic. This may be an issue for high traffic websites.
Lokalise
Lokalise is an AI first, developer friendly TMS platform that is widely used for app localization, which is where most EPD led companies start out.
Best for EPD led startups and growth companies that are experimenting with localization.
Strengths
- Developer friendly TMS that started out with a focus on app localization
- Low cost self-service entry points, popular with EPD led companies that start with app localization as a proof of concept
Weaknesses
- Integrations with other platforms are not as extensive. They have improved in recent years, but are limited in depth compared to platforms like Phrase and Smartling
- The user interface is a bit dated. Not a big deal, but marketing and design teams may find this off putting
Cost
Lokalise, like CrowdIn and Transifex, also offers a free plan as well as low cost self-service plans, which likewise positions it for use by early stage companies whose budget is limited.
Lilt
Lilt was one of the first TMS platforms to really lean into AI translation and focuses on companies that have high volumes of content to be translated. They primarily focus on large enterprise and public sector entities, which makes them a bit of an outlier for our target audience (startups and growth stage companies), but we are including them for completeness.
Best for enterprise and public sector entities with large volumes of material to be translated via AI first workflows
Strengths
- Solid AI first architecture with support for human in the loop and fully automated workflows
- Integrations with most systems that matter (CMS, help center, etc)
Weaknesses
- Not focused on small to mid sized companies
- No low cost self-service entry point, every plan involves talking to a sales rep (not a match for EPD led companies)
Cost
Fixed and variable costs are negotiated on a case by case basis. There is no “rate card” per se.
Phrase
Phrase is a European based company which was formed via the merger of MemSource and PhraseApp. MemSource focused on long form content localization, while PhraseApp focused on app localization. Combined they offer a comprehensive range of TMS solutions for many use cases.
Best for scaling companies that are committed to localizing across user touchpoints.
Strengths
- Broad range of AI first translation management workflows for use cases from app localization to long form content
- Solid integrations with popular CMS platforms, help center/CS, etc
- Suitable for EPD led and marketing or design led companies
Weaknesses
- No low cost, self-service entry points, more upfront commitment is required
- Frankenstein product. The two products are not fully integrated, which results in duplication of effort across projects
Cost
Phrase, like Smartling, tends to focus more on larger companies and growth companies that are committed to localization and global expansion. They do offer self-service plans starting at $1,250 per month, which is low in the overall scheme of things, but may be too rich for an early stage company.
Smartling
Smartling was founded in 2008 by Jack Welde, and was among the first cloud based TMS platforms. Until then, most of these systems were on premise systems and frankly were a nightmare to deal with. Up to that point most translation technology was focused on enabling translation agencies to work efficiently. They were not tech forward companies and their products showed it.
Best for scaling companies and enterprises that are committed to localizing across user touch points.
Strengths
- Integrations : Smartling is integrated with most other platforms that matter. They’ve done this so you don’t have to
- Longevity : they’ve been around longer than almost everybody and have had time to sort out most bugs and edge cases
- Operational support : their professional services unit can handle most routine issues and has high service levels
Weaknesses
- No DIY : everything involves talking to an account rep. There is no self-service product. They’ve missed some huge opportunities like OpenAI because of this!
- Expensive : Smartling is definitely one of the more expensive TMS systems, but in the overall context of global expansion, the spend on translation tech is a rounding error
- Conflicts with LSPs : they have their own language services division and aggressively promote it, which creates conflicts with LSPs (language service providers)
Cost
Smartling is definitely focus on the premium end of the market. They offer a solid platform with well maintained integrations, but they are not cheap compared to other vendors. As the saying goes, “Nobody got fired for buying IBM.” They don’t offer low cost, self service options and instead focus on companies that are already committed to localization. Annual costs will be $50,000 and up depending on content volume.
Transifex (now part of XTM International)
Transifex was one of the first cloud based TMS platforms along with Smartling. They started out as a developer focused platform (the founder Dimitris Glezos is a developer). They were recently acquired by XTM International to round out their service offerings.
Best for EPD led companies that are starting out with app localization.
Strengths
- Developer friendly, AI first translation platform which is a good match for EPD led teams
- Upgrade path to XTM service offerings for more complex localization needs
- Low cost of entry for early stage companies and proof of concept projects
Weaknesses
- The company is mostly focused on EPD led customers, so if you are planning to translate many customer touch points (CMS, help center, etc), you may need to build and maintain your own integrations.
Cost
Transifex, like CrowdIn, LocalizeJS and Lokalise, offers low cost, self service plans starting at $125 per month.
Discussion
Each system is well suited for specific use cases, so there isn’t a clear winner across company domains. Which platform you choose will depend on the following:
Why shouldn’t we build our own TMS? It’s just a database of strings isn’t it?
I mention this because I hear this a lot from engineering led teams. While that’s superficially true, there are a lot of edge cases in text processing and localization. Some of these companies have been operating for almost 20 years. Unless you are Google or Facebook, you definitely do not want to re-invent this wheel. It’s easier and much faster to write a check and let someone else worry about this.
How much engineering support do you have?
EPD led companies will do well with self-service tools like CrowdIn, Lokalise and Transifex. Marketing and design led companies and teams will be best served by tools like Smartling and Phrase.
Are you committed to localization or experimenting?
If you are making a multi-year commitment to global expansion, definitely look at enterprise focused tools including Lilt, Phrase and Smartling. If you are experimenting, look at developer centric products like CrowdIn, Lokalise, and Transifex.
Do you have legacy assets with tech debt?
If so (and you probably do), definitely look at LocalizeJS as a solution for localizing them quickly and inexpensively.
Do you have a centralized localization team or is it dispersed across the company?
Early stage companies often start localization in an ad hoc manner with different teams owning parts of it. It is generally good to have a centralized team that owns localization infrastructure and service vendor relationships. If the function is centralized, it is best to look for a TMS that has mature integrations with other platforms. If, on the other hand, you are on a lean EPD team that is building out a proof of concept, developer focused TMS platforms may be a better fit, while marketing and customer success teams may use a TMS that is better suited for long form content.
Do you have a personal favorite?
I’ve worked with all of these platforms (often the choice of TMS had been made before I arrived). They all perform similar functions and do well at them. One thing I have experienced throughout my career is that localization teams are almost always under-resourced, especially with engineering and operational support. Because of this, I favor Phrase and Smartling because most of their integrations work out of the box and we don’t get blocked by lack of engineering support. CrowdIn also has a lot of integrations and is a good option for EPD led teams.
What about vertically integrated systems?
There are several companies that offer vertically integrated systems that combine technology and translation services. Transperfect is a leader in this category. I strongly recommend against using these, and instead choose the translation technology platform separately from language services to avoid vendor lock in.
One of the nice things about translation management systems is that you can have multiple service providers working within that environment. It is common for service providers performance to vary by language or content type, so being able to route work to the best vendors is helpful. If you are leasing technology from your service provider, you are trapped in a hostage situation.
Not a good place to be, even if you develop Stockholm Syndrome and fall in love with your captor.
How much will these systems cost at scale?
Most of these systems are priced based on the number of words hosted, or in the case of LocalizeJS, based on the number of pages served. I generally tell clients that translation tech will usually account for around 10% of their overall localization spend because most of the spend is on services that these systems mediate. The most important price difference is the entry cost, which is where CrowdIn, LocalizeJS, Lokalise and Transifex have an edge.
On this topic, it is good to understand that TMS platforms are a niche market with a finite number of customers. Integrations with other platforms like CMS are expensive to maintain. These two factors explain the high pricing relative to most SaaS platforms which are based on per-user or per-agent pricing.
Get In Touch
Even within lean teams and companies, the dependencies between TMS platforms and other platforms (CMS, help center, lifecycle comms, etc) can be complex. We can help you navigate this decision making process and insure that you are buying systems and infrastructure than scales with your company.
Reach out to brian@loctechpartners.com
Related Reading
Dynamic Translation : Building For Runtime Localization
How Much Should I Budget For Localization
Choosing A Content Management System (CMS)