As with any multifunctional platform, it can be difficult for new users to get started with and make the most out of the many features available in Smartcat.
When we sat down with Viktor Kuryshev, owner of GTMK, a Volgograd-based translation agency practically built on Smartcat, we knew he would have some practical tips and ideas to share with us.
Viktor’s proficiency in finding suppliers via Smartcat’s marketplace caught our eye so we asked him to share the process he and his colleagues have developed over the years of successfully using Smartcat to tackle projects of various size, scope, and complexity.
Now, a word to the wise!
Table of Contents:
Like many other agencies, we migrated to Smartcat, along with our in-house translators, after failing to implement Trados, the behemoth of the translation industry which turned out to be too expensive and in many ways outdated. Back then we were quite small, and without significant investment, building a robust and transparent workflow in Trados didn’t seem realistic.
As the company kept growing, we started offering new language pairs and subject areas to our clients, which meant we needed more and more specialists. We were able to find many of them on Smartcat’s Marketplace but also brought some from outside and successfully added them to our account. That being said, we didn’t stop these linguists from opening up their new Smartcat profile to other customers on the platform.
Below are some of the proven methods I discovered and refined to make our vendor hiring and management process as seamless and fruitful as possible.
Imagine you have a large document to translate, the deadlines are tight, and you don’t have any linguists for the job. Finding qualified translators in a very short time is a challenge, but I’ve been down that path before and developed a system to tackle the usual challenges.
Search and selection
First off, I head over to Smartcat’s Marketplace and set the appropriate filters, such as the source and target language, service, subject, etc. and hit “Search”. Next, I sort the linguists by rate.
When looking at individual profiles, I pay close attention to the number of words the linguist has translated. If the word count is north of 300k and there is some positive feedback from previous clients, that’s our candidate. If there are no reviews in the profile, however, that might actually be a good sign as at least there are no complaints — it’s possible that the linguist has regular happy clients, and they just don’t care about getting feedback on their profile.
But any reviews are definitely worth checking out. The most important points in a review for me are 1) who left the review and 2) the language pair of the project. Having a lot of positive reviews for translations from Russian into English is no guarantee that the translator will be able to provide equally high-quality services in the reverse pair (English to Russian). However, when dealing with rare language pairs and tight deadlines, this might not be a big issue.
Profile descriptions also deserve to be studied carefully. Pro-tip: Preview the description text by hovering over the linguist’s name to save time. As I scan the profile for details, the linguist’s experience is what matters the most to me. Additional information, such as examples of previous work, is of less importance when searching in urgency mode.
Now, let’s look at the numbers. Say I need three translators to get the project done on time. To make sure I have a broad enough selection of candidates to choose from, I’ll send out invitations to around 100 people. About half of them will respond, with 10–15% responding within an hour and the rest taking up to three days. I’ll have to negotiate the terms with about a third before they can start, while two-thirds will get to work immediately. Ultimately, it all will boil down to 15–20 linguists who are available and meet our quality standards.
Sadly, some candidates ignore the text with important details about the project that I always include in my invitations. To avoid miscommunication, I make sure everyone I work with is aware of our requirements and preferences by sending these important details directly via live chat. Below are some examples of accompanying text I add to my invitations:
“If you have any questions, no matter how silly or unprofessional they may seem, please don’t hesitate to ask”.
“If for any reason you are unsure about how to translate a segment correctly, please skip that segment and report its number. We’ll take care of those segments in-house.”
“If the document has been split into segments incorrectly, please DO NOT insert the source text into the target segment. Simply report the issue and carry on”.
One video is worth a thousand words! So, I’ve recorded a series of narrated screencasts demonstrating how to use Smartcat’s Editor and its key features, such as translation memories, glossaries, machine translation, concordance searches, etc. I share the link to those videos with any new linguists who are not familiar with Smartcat.
Once we’re done with the preparations, it’s time to move on to the testing part. By assigning a standard 150–200-word text to the linguist, I am able to assess the quality of their translations. This takes time but definitely pays off. Oh, and our test jobs are always paid, whether the candidate is ultimately hired or not.
Once the linguist has completed the translation, I assign an in-house editor to review it or do it myself. If I’m satisfied with the quality, I add the linguist to My Team, ask them about the volume they’re willing to take on, and finally assign the actual task to them.
All translations done by new team members are checked by our in-house editor. By “new” I mean any translators who have not yet worked on 50–100k words with us.
Once the editor has finished checking the document, I download the original translation and its edited version as separate bilingual DOCX files, compare the two documents using MS Word to highlight the differences, and export the resulting document to a PDF file. I then send that file to the linguist so that they can review the edits made and reflect upon them.
At the 50–100k-word threshold, the linguist is usually qualified and reliable enough to handle our projects with the integrity and care we require. Regardless of the linguist’s proven record, all translations are refined by a proofreader to ensure accurate and polished results.
Forming a translator pool
If there’s a large project coming up but there’s no rush or if we have recurring projects in a specific language pair or subject, I take the time to source the right candidates beforehand, which also gives me the opportunity to add new members to my talent pool for future projects.
First, the editor prepares a sample text based on our previous projects to be used as a test. I then contact the Smartcat support team asking that they post the test on the platform without the mailout option.
At the same time, I draft an announcement saying that we’re on the lookout for new translators which I then publish on various online job boards and community platforms, both paid and free.
Once applications start rolling in, the process for communicating with the candidates is somewhat similar to that above — I share the details of the project and, if the candidate is not a Smartcat user, explain that the work will be carried out in Smartcat. I also explain that the candidate will need to complete a test and give them the link to the job posting published on Smartcat’s job board. For those not familiar with Smartcat I provide a link to a step-by-step video tutorial of how to sign up to Smartcat, apply to the job, and complete the test assignment using the Editor.
If the project start date is fast approaching but the test results are not sufficient to select enough qualified candidates, I create a test project with a document sample and invite anyone from the Marketplace with the relevant language pair and rates matching our budget. My go-to approach to rates involves multiplying the highest acceptable rate by two, as some translators agree to lower their rates for large projects or the opportunity to get regular jobs.
Next, I contact those who have accepted my invitations, share the details of the project, and give them the link to the test job mentioning that if they have already completed it, there’s no need to do it again.
If there is no rush, I also post a job offer in Smartcat’s “Freelance jobs” section and then the process goes in the same order as described above.
It also makes sense to ask Smartcat’s support team to keep the job offer posted for a longer period. This way, when we’re found by newly registered candidates, we can run them through our procedure and include the best ones in our roster even if we’re not actively seeking new additions to the team at the moment.
Validation, step 1
Once a certain number of linguists have completed the test in Smartcat, I download the results in a batch. Note that Smartcat only allows downloading the test results in 50-item batches.
Next, I send the batch to the editor if there is one available for that language pair. Otherwise, I reach out to the Smartcat support team for assistance which is a pricier option but well worth considering if you need to find a qualified linguist fast.
I dismiss all the linguists whose test results I’ve already downloaded to be able to tell those who have completed the test from those who haven’t.
Finally, the editor reviews the test results and creates a spreadsheet in which each candidate is graded on a scale from 1 to 10. Translations rated 1 require way too much editing to be feasible while those rated 10 can be described as nothing short of perfect. The spreadsheet also contains a dedicated column for comments on errors found which makes it possible to assess quality more objectively. For example, if a candidate rated 9 made semantic errors, that’s a no from us. On the other hand, if the errors are merely stylistic, we’ll likely add that linguist to our talent pool and start giving them occasional tasks, gradually increasing the volume over time. All candidates rated 10, as well as some 9ers, then progress to the next stage.
Validation, step 2
This part goes in the same manner as in the urgent scenario. I add the successful candidates to My Team, ask them to complete the test, and provide them information about our requirements, as well as video materials.
As with any efforts that involve dealing with people, vendor hiring cannot be fully automated. However, certain steps in the process can and should be simplified. That’s where Smartcat steps in, allowing translation companies as well as end customers to find and hire language service professionals — both freelancers and LSPs — from among the 250,000 suppliers registered on our open Marketplace.
Do you have any helpful tips and tricks you’d like to share with us? Just leave a comment below!