One of the lessons I’ve learned when I started selling things (some time ago), is that you need to encourage your clients to show they’re serious, just like they expect you. Being serious means they can spend money.
So, how do you reconcile the constant requests for “free test translations” with the need to get some minimal commitment from the client?
When new (potential) clients approach us and ask for test translations, our goal is to build engagement. Engagement is a two-sided process. On our side, we do some work for free and on the client’s side, they evaluate this work, give feedback and would consider us for their translation work.
I’m thinking that we should have an official, published policy for test translations and I want your help to get it right.
How to Get Test Translations
In my opinion, the best way for giving test translations is from the project chat.
Clients invite translators to apply for their projects. This is already the beginning of a fruitful engagement. As translators, you need to spend time responding to these invitations, so translating a paragraph would not be any more effort.
We can encourage clients to send a test paragraph when interviewing translators. This would give clients what they want and would also give us what we need (direct interaction with the client).
Length of Test Translation
The second question is how long should test translations be? Of course, they need to be meaningful, but they shouldn’t turn into projects. And, we don’t want to be doing test translations the whole day. It’s supposed to show clients how we work, not do the whole job as a test.
What would you recommend for maximum test for test translations?
42 responses to “Our Policy on Test Translations”
usually, in romania, clients send you about two pages of such test, which is quite unreasonable; and if you “dare” to tell them that it is too long and that you do not intend to do the whole translation of a document for free, they send it to someone else, who is so eager to get some work that they do it. for free. and they never get any feedback nor get translation jobs from those clients. it’s sad
Well, that’s exactly the situation I’m trying to avoid here. In your opinion, what is reasonable length for test translation?
In answer to the first question; when I first started doing translation work I was often asked to do a ‘test’ translation for customers/clients. I had read somewhere that this never lead to anything, and that these people were just fishing for some free work. Nothing could be further from the truth. Almost every time I have done a test, it has lead to serious work. I totally agree that doing a sample translation is a great idea; it gives the translator the opportunity to show off their work, and the customer a chance to judge the quality of their translation skills.
In answer to the second question; I think that in the case of ICL, I would be willing to take a test translation of up to 300 source words. I think most of us can work through 300 words quite quickly, and I like to think that people looking for a smaple are generally honestly looking for a good translator, and not out to trick us.
Question for Amir: is there any way to allow potential customers to view examples of ‘before and after’ work already done for existing customers (if they agree, and it’s not in violation of any confidentiality agreement). This would allow customers to view work actually done on this website, it’s verifiable, done by a specific translator, but doesn’t creating extra work 🙂
Being able to show “before and after” work is a good idea, but I would keep it only to material which is publicly available and not something that was sent to you privately.
We contact clients after work completes and ask their permission to showcase their work. I’ll see if we can also include the nicknames of the translators who did the work in the showcase.
Let’s start by offering clients to send a short test text and see what kind of response we’re getting from it. Then, we decide how to move forward.
Maybe if we all, translators at icanlocalize, could all have some official credentials/official diploma/official recognition, then we wouldn’t have to do these tests. The fact we have them would be enough of a proof for the client that we can handle the job.
It could be a bit time consuming in the first place to have all translators with a professionnal recognition, but then it would be a time saver as you’ll never have to prove again to your clients that your translators are good enough.
I know some companies have this policy. I mean it’s aso the same with any professionnal field, you don’t ask a surgeon to do a test on you before going for surgery… (even if you can argue that it should be a good idea in some cases!)
Translation is not only about formal qualifications. Having a degree in translation is a necessary step for any translator signing up with ICanLocalize.
However, when clients choose their translators, they often consider writing style and previous background in their field. In most cases, this is what the test translation is for.
It’s not a test for your language skills, but to see that they simply like the way you write.
Thank you very much for a newsletter. As for our company – AUM Translation Services – we always do a test tarnslation before starting a big project with a new client. The volume of test translation is usually about 250 words. The maximum volume of test translation is 500 words. I think 250 words is quite enough to understand if it is a good translator or not.
Great. I’m glad to see we’re all in agreement about this.
That might have been a mistake somewhere as I don’t have a degree in translation…
I understand your point Amir when it’s about literary translation, but most of the time we are dealing with iPhone, iPad Apps, quite technical material and the challenge is more to reduce the text to a few basic words.
So maybe tests for some fields of translation and not in other fields?
You’re very correct.
For iPhone and other SW localization, we don’t normally get requests for test translation. We get that a lot for website and document translation.
I also heard from fellow translators who were asked to do test translations for companies in the past. They have had lots of bad experiences and said that most of the ‘clients’ asked them to do test translations and never sent feedback.
Maybe it is worth a try though and we should see how it turns out. I think 300 words can be quite a lot if it is a very specific text that requires research. I would suggest max. 200 words.
Also, I think the client should only ask one translator to do a test translation. As most translators work professionally and deliver good work, it would be a waste of time for x translators, as only one translator will get the work.
When we recruit translators from Proz, we often send short texts as sample translation. Normally, this is really short, like one paragraph.
It helps us a lot. We can quickly see which translator writes more fluently and naturally.
Almost all ‘test’ translations come back with correct language, but it’s a matter of getting the right translator for different kinds of material.
If it is only one short paragraph that should be fine!
Also, how can they assess themselves the quality of the test? I’m asking genuinely this question as it seems surprising. I guess you do it, but they should be warned that comparing the test to some machine translation is not a good idea (maybe by handing them some funny examples). You meet daily people who think that machine translation are great (because they don’t work in the field) and are very surprised when you tell them it’s good yes, to get an idea of a document in a language you don’t know, but that it is not good enough at a professional level.
I suppose they all don’t have the same way to assess, maybe some know the language, through a friend, through machine translation… I can’t imagine they pay another company to assess the result of the test!?
Maybe it would be a good question to ask clients how they do that?
Many times, they would either have staff that speaks that language, or send it to someone who does.
The fact that clients have people speaking a different language in the office doesn’t mean they want to use them as translators.
Who’s watches the watchmen, right?
We could even reduce test translations to 100-150 words. That would keep off clients who are not interested in actually hire a translator.
Test translations should be no more than 200 words, and although that’s not the all the time, perhaps the client should make a small payment for the time the translator has to spend on the test translation – which can be regarded as a ‘test payment’ for the translator to see if the client is really serious or just taking advantage 🙂
I am also working for another online translation website and that is exactly what happened. The client picked 10 potential translators and sent out a small paragraph to be translated, but he did pay for it for a special price that he negotiated with the admin(I think 50% less of the usual rate) as a compensation.
I definitely understand why serious clients would want to do a test translation, but two pages of test translation would be taking advantage and I doubt that I’d take part.
But anything between 100 and 200 words is fine by me and I would be more than happy to do the test.
Great. +1 vote for adding it quickly to our system.
I think a short test (150-300 words) doesn’t take us much time, we can see what kind of translation is and we can decide to do it or not, according to our timetable.
I find the idea to show some translations of ours (from the same/similar subject) is also good and professional. The client can see we have already experience in this field.
Unfortunately it’s very difficult -sometimes- to recognize the “good and fair” client!!!
And all of us know that speaking a language, even very well, doesn’t make us translators… it’s job which take many years of perfecting, like most of others professions. I know many people who speak very well different languages and wouldn’t be translators. The problem sometimes is that everybody think he/she is an expert when it comes to language. I can do electric cabling but I’m not an electrician. Maybe it’s just it’s a young industry without enough national or international guidelines yet. Like a Blue Card, a Builder card, I don’t know the equivalent in each country. Because when you order a translating job from a reputable company with reputable translators and pay also for proofreading, I mean we can in the end argue about a few words, but as Britta said most translators know their job and in the end a good job will be delivered.
Yes, I understand it might an issue for you to find the right translators for the right jobs Amir ; I think even in my translating work I am still discovering what are the right jobs for me and my skills, the jobs I should do and the ones I shouldn’t.
Reasonable tests are always welcome, but I think it’s necessary to charge a little fee to help filter the serious clients from cheaters (you know, a lot of people out there get whole projects translated simply sending parts of it as tests to several translators), especially because most of cheaters are not willing to pay even a dime for a test, while the serious ones don’t make a big deal about it.
But the problem is, the same client may be sending invitations for a dozen translators, so if we ask for payment, it’s going to end up a lot.
I prefer to limit the test translations to pretty short texts (say, 300 characters) and encourage clients to pay if they like a certain translator and want more translated for evaluation.
Well, I think just by reading our profiles and other clients’ recommendations will provide the client with a good picture of which translator would be more suitable for the project, so instead of inviting a dozen of translators they would select just a couple for testing.
One more thought, maybe a convenient way for adding tests to the system is making them optional, therefore on each translator’s profile would show if they accept tests for projects with a X number of words, maximum words per test, if they charge a fee or not, and so on.
Hi everybody: my opinion on that matter is that ICanLocalize is a company
that sells translation and quality. The client buys ICanLocalize’s service. So what’s the
point on a translator contest? ICanLocalize should be responsible for the
quality of the translators and the translations.
ICanLocalize is a serious company, we are serious translators. The client should
not disagree with a professional staff like that. We do not argue about the
quality of a source text sent in by the client, even though very many of them
are very bad. We are as respectful of them as the client should be of ICanLocalize’s
We must not translate what the client understand but what we believe is correct in our
language. Who is going to check the quality of a translation intended for Argentina:
a Spanish (from Spain) speaker living in Holland? I guess the style of a translator
can be assessed by examples in our profiles and not by ad-hoc tests.
A test is a project in itself. The client must pay for it – even if we take into consideration that
very many of the translations at ICanLocalize are very short and may represent
the whole test. A test is time spent within ICanLocalize system. It has a cost for everybody.
My main point is that serious companies do not go around proving they are good; they
are responsible for it. Clients believe in what they say.
Thanks for the opportunity to speak out my opinion.
You’re right about everything, but clients are not mind-readers.
There isn’t a single translation service that would say:
“We employ bad translators and don’t care about our work”
Everyone claims they do a good job. We should be able to demonstrate it.
I think that doing a translation of maximum 350 words is in the upper limit of being reasonable.
The second aspect to be considered should be the field of specialization, so that the sample translation requests, reach all translators who can handle the subject with experience and proven professional ability.
Right again. In the form where we offer sample translation, we also indicate the fields of expertise. You can see that when you apply to the project and can decide if it’s good for you or not.
It was interesting to read your post. In times when virtually every minute may matter a lot, spending extra time for some “test translation” appears to be quite a dubious deal. I can compare such a request with the following message: [to a designer, my own experience] “Could you please send us two or three variants of our logo design, so that we can choose one and pay you afterwards?”
Such a request provokes only one reasonable explanation: a client that does not trust you can neither expect trust. I may sound radical, but when I reveal my identity, school certificate and university degree diploma justifying my profession and expertise, and offer my services I would be glad to be treated adequately.
Moreover, ICanLocalize administration will have to remove from the site the request to do practice projects, because it will be extra since all clients would want a test translation from applicants.
Nontheless, I am ready to do test translation if, for example, at the request of one ICanLocalize staff will provide their own sample text to demonstrate the (potential) client the expertise of a selected translator. This is only due to the fact that I trust these people and know they won’t fail me.
Just to explain, our practice projects have nothing to do with language skills.
The purpose of these practice projects is to train you on the tools. The practice project was especially built to walk you through all the features in our translation software. In most projects, we catch small usage errors and need to explain how to do things right (different errors, here and there).
I’m pretty sure you don’t want to ask us technical questions about using the translation system while you’re doing your first live project for a paying client and are under a tight schedule.
I agree with everything else.
Personally I thought practice projects were a sort of evaluation, which could be quite right. I mean that along with learning how to use TA only, it could be nice to use it as entry test. New applicant completes this test and one of ICanLocalize reviewers evaluates it. Then no client would doubt the qualification of translators of ICanLocalize, because they would be able to see the results of the test being obligatory requirement.
I would strongly encourage free test translation, so that the client ensures the quality and decide whether to proceed or not. I would ask you to help me in doing practice project, so that I start working with you.
I think I have the solution.
Client X has a project of, say, 5,000 source words to be localized in 4 languages. He can ask for a test translation of max 150 words taken from the project and will pay for 4 translations as usual.
35 different translators decide to deliver their test translations. Then the client picks up the best test for each language and the 4 winners get paid by ICL. If the project is confirmed by the client, that is 5,000 source words, it will be automatically assigned to the winning translator for that language.
Simple as that.
For iPhone apps, I would not allow any test translation, because projects are too small, as a general rule.
That sounds very reasonable to me.
I like this suggestion. Then the translators can choose if they want to do the test translation or not and the client gets the best possible translator for that specific project.
P.S.: Of course the translator would be free to work at his/her own rate.
I think requiring a translator to do a test of more than 250 words is unnecessary. You can get an idea of someone’s skills and writing style in a translation of that length.
sry, no way.
a) test translations are usually, if at all necessary, up to 150 words at most.
b) they are necessary only if a fresh translator has no sample translations and no references yet while a new business partnership is starting with the agency wanting to evaluate a translator’s skills
c) for all other cases (translator has samples, references/feedback in their profiles), there is absolutely no need for test translations, we are not bored people searching for a spare time entertainment. We are doing this job for a living, it’s our profession if I dare to remind. I am definitely not offering test translations here at ICL, where a client has anything he wants to evaluate an LSP’s quality: feedbacks, profile, possibility for contacting and requesting samples etc.
No way. Whoever wishes to offer his work for free – good luck with this. I hope I am not the only one here saying no to such expectations.
short complement note – please let’s not forget that a client can always open a “test translation” job in order to plan a larger project with a click on a button. a simple smaller order, no business risk for the client and yes, it has to be paid like any other work in the world.