An open secret about legal translation services is that they are subject to not one, but rather three kinds of budgets. Each of these budgets is really a kind of cost that applies to the translation and the further each factor is constrained, the lower the quality of the translation. The first kind of budget is translation provider consolidation: some companies work with one intermediary who does everything, and that intermediary has a network of a dozen other intermediaries to which it outsources. Consolidating relationships using supply chain methods usually accounts for 70% of the translation budget. The second budget is timeline; the deadlines in translation are the shortest of almost any white collar professional services industry, which results in qualified translators being unavailable and existing legal translators having to rush. Finally, price is a major budgetary constraint but not as significant as the first two; the reason is most translation agencies try to spend as much of their budget as possible on timeline and consolidation while minimizing the money spent on actual translations. Most legal translation clients are unaware of this and, as a result, unwittingly choose a positioning that minimizes translation quality.
Consolidation
For most law firms, the cost of having a single project manager consolidate relationships with multiple translators is about 65% of the total translation price. As supply chain management costs go, this is not at all excessive; if you look at the marketing costs for things like carrots, spending 70-80% on marketing is normal. In the context of agricultural products, “marketing cost” refers to the cost of bringing something to market and is rooted in the etymology we still see today with farm-to-market roads. According to Nimdzi and public translation companies’ financial reports, the average blended net profit margin among translation companies is 12%, while the gross margin is about 40%. Most translation companies do translations for other translation companies, so one company is doing the marketing with a 40% margin whereas the second company is staffing translators, and, in Chinese translations, this is also about a 40% margin with a 12% net profit margin. Small translation company owners not using a shareholder model doing this make on average about $60,000 annually. With this dual-layer supply chain, about 65% of costs are spent on finding linguists and signing them up, then keeping them from cheating the anonymous client.
Most translation companies are intermediaries because a given client’s demand will be spread very thinly over a very large number of specialties and hiring full-time staff for that demand is not feasible. To compare it to the legal field, the work done by intermediaries is very similar to contract attorney staffing companies, and a lot of big translation companies are now even getting into the contract attorney staffing business as a result. One aspect of this process that’s very different is that there are about 50,000 translation companies devoted to registering a pool of over 900,000 linguists, and the same linguists’ information will have to be registered and put into databases hundreds of times. The sourcing aspect of the job, along with sending out salespeople to pitch to clients, accounts for most of the gross margin costs and is usually accounted for as staff costs. An advice book for new translators says that it is normal for a translator to spend many years where 70% of their time is spent on simply filling out forms and signing up for these agencies.
In this milieu of translation vendors, there are generally four main types of translation providers. The first is freelance translators, who account for the vast majority of workers; statistically, they are mostly casual workers who also work as housewives from home on a laptop. Second are professional translation firms, accounting for about 8% of the industry, who are focused on working directly for clients and providing high-quality service. In the next category are Single Language Vendors (also called SLVs or SLSPs), who mostly sell to the fourth category, Multi Language Vendors (MLVs). At minimum, an MLV will probably have a list of 30 SLVs.
The industry guidance on these MLV translation companies says “quality is not important” and, while that may be true for most of them, some boutique firms focus on quality. Of course, they all tell clients that they provide top quality though, in reality, most focus on hiring from the freelancer-housewife category of casual workers, since most clients cannot tell whether the translation is accurate according to leading industry manuals. A Coursera course for professionals at these companies advises that choosing the cheapest workers is the only way to make up for the consolidation cost of the business model. Thus, most of the cost is paid for in a concept called “quality fade” described below, also called “skimpflation” by NPR.
Case Study 1
Any law firm or corporation purchasing legal translation services should run a cost-benefit analysis on the cost of consolidation. For example, I know several China practice attorneys with pretty large work volumes in complex corporate compliance and litigation that use MLVs. Their annual volume is about 3,000,000 Chinese characters at a rate of $.20USD/character, which costs $600,000 per year and $390,000 of this is supply chain costs. On the SLV end, this is technically a job for 8 people… assuming you only pay those 8 people $20,000 a year. So, they reduce it to 3 people or full-time equivalents, who are using free machine translation and AI rewriting tools to do the work and checking to see if anything would alert a client to a problem.
The result is that the client is paying a $130,000 headhunter fee per translator, and having all of its intellectual property sold to companies like Google and Baidu Inc in exchange for the free AI services used by the translators to cover the hidden costs. Usually, law firms do not run the math and realize that its linguists are, in theory, being paid an unrealistic $20,000 per year. Often, they have bilingual paralegals telling them that the translation quality being produced by the AI is great. The problem is that the AI companies also produce the dictionaries (also written by AI these days!) the paralegals rely on, thus creating a false feedback loop that tricks internal layperson quality controllers.
An alternative for the client, and one some big law firms with high-stakes China-related matters actually use, is to require the company to use certified translators who have sworn to a code of ethics. The result is that there isn’t huge skimpflation. Generally, a translation company is going to have to charge twice as much to get those translators onboard, so the lawyers are now paying $260,000 headhunting fees per certified translator. The translation companies in this case would just be calling up a bunch of people on the ATA Certified translator directory to find people who other experienced translators think are competent.
This isn’t really the fault of the translation company. They have no idea what the law firm is doing or why, just a series of purchase orders that fall out of the sky which they quickly fill. Only the law firm can predict how much headhunting fees are going to be. If those headhunting fees are used to pay for genuinely qualified translators with a strong sense of professionalism, the results for the law firm would be far better.
Bottom line for lawyers: do a cost-benefit analysis on your project to determine if the portion of the budget allocated to vendor consolidation really makes a lot of sense.
Case Study 2
In a second case study for the same law firm above, a compliance matter reveals that several employees of a Telecoms company deployed to South Africa are writing e-mails in Tsonga. If you were to ask me for help with this, the best thing I could probably do is to get in touch with an SLV in South Africa that routinely handles these kinds of requests and have them handle the request. 800 words by an SLV might require around $40 in headhunting costs for a sworn translator of Tsonga. The SLV would likely spend more money tracking down and calling up an accredited Tsonga translator than it would profit. In this case, the SLV, or even consolidating with an MLV to find someone to help with the e-mails, is a good investment in consolidation.
A lot of lawyers I worked with in the FCPA/AML fields like to drop names of big translation companies as a sign of quality assurance. I would caution these lawyers to read what training manuals and industry consultants have to say, that “quality doesn’t matter” or “always look for the cheapest option.” Insiders’ reviews on Glassdoor accuse a lot of these companies of misleading clients. Intermediaries can be your friend, but they follow no professional ethics code like the certified and sworn translators’ program, you need to know how to collaborate effectively to minimize risks.
Timeline
Translation is extremely time intensive; expect about an hour of labor for every 300 words to achieve what professionals consider adequate quality. Most translators do not charge by the hour, and there is a good reason for that: clients are usually in a big hurry and try to get translators to spend less time! This leads me to the second big cost factor in translation, which is speed. A lot of translation agencies charge rush fees, though actual translators usually do not. The reason an agency charges rush fees is that calling up a dozen translators to find one who is available is really time-consuming. In a typical situation, the client also pays a pretty big “quality fade” premium as a result.
Anyone familiar with contract attorney staffing methodologies should be familiar with how this approach works. A staffing firm will make a list of workers ranked from best to worst and start calling them in the order of top to bottom. A translation company will also do something similar within a single appropriate price bracket, i.e., the cheapest that won’t cause a client to notice quality defects. A list with, for example, 15 translators on it may have 12 who have already been signed up in this manner. So, by the time you get to #13, you generally get a fairly unreliable translator with significant problems that most people don’t want to hire. A major and key difference here is that legal staffing agencies use licensed attorneys, not just random people off the street. A rush translation project often does involve a project manager just grabbing random people off the street, who have no training or credentials at all.
Suppose you ask internal staff to do the work on a tight timeline: what happens then? I’ve checked numerous companies’ internal translation results, and what usually happens is that the internal staff just copy and paste Google Translate without revising it, despite the fact that the machine translation engine is still fantastically unreliable and has caused numerous publicized disasters. Lots of managers, under pressure to cut costs, will instruct internal translators to just copy machine translation without verifying whether it’s accurate or reliable. In China in particular, I haven’t seen internal staff at companies not blindly copy Google Translate, often they don’t even read what comes out. Fascinatingly, American company managers just think they are producing incoherent Chinglish, and don’t realize that these staff are just copying from Google Translate before going back to play with their cell phones.
A professional translator, as opposed to internal staff or freelancers, will usually warn against unrealistic timelines. They can of course “rush” a translation, drop everything, and spend half as much time on it, but ethics requirements mandate communicating the risks to the client and having informed consent. In this case, half of the translation budget being spent could be allocated to getting it done fast. The price customarily doesn’t change in these circumstances but spending half the time means half the human input, and therefore half the value from the translation. If your ROI from the spending was 300% before, it might be 150% afterwards. For a lawyer charging $600/hour, relying on a low-quality translation means the value delivered to the client could be cut in half or even zeroed out.
The Bottom Line: Lots of people casually sacrifice a lot of value without doing a cost-benefit analysis on rushing the legal translation. Getting work done quickly in many circumstances often does have huge value, and this can outweigh the decline in translation value.
Price
By “price” I am referring to the price charged by the actual translator, after all of the various headhunting and supply chain management costs are taken into account. The quality, and therefore value, of the translation varies with the price, and for fields such as legal and patents, the return on investment in the translation at low price levels will be negative. That is to say, if lawyers charge a client $20,000 to give erroneous legal advice based on a sham translation, then the translation just caused the client to flush $20,000 down the drain. A study was done on Netflix where bad translation quality was shown to result in 70% of foreign viewers turning off the show. That is also a case where ROI can be negative. Apparently, Netflix’s outsourcers haggled subtitle translators down to $400 for some TV series scripts, and they responded by using machine translation with AI editors to send back faked work.
In some fields, such as search engine optimization, the translation can get extremely cheap before you lose any money. In fact, lots of big translation companies these days are even refocusing on being content marketing providers rather than translation companies.
Price affects the value of a legal translation for two reasons that have to do with training and value. To start with the less obvious factor, learning legal translation well enough to pass certifying examinations offered by the ATA in America or CIOL in the UK takes many years of effort and dedication. The translator will have to read numerous books and do a lot of practice. However, the majority of people getting into translation are looking for a relaxed, work-at-home job. It’s a casual job for them not unlike what Uber offers drivers. The group of casual workers generally does not engage in a lot of training or study activity to build the skills to pass certification. The difficulty of passing certification is about the same level as a law degree or maybe a medical license: many thousands of hours of intensive study are needed to develop those skills. Most translators will have done less than 300-400 hours. I’ve surveyed hundreds of Chinese translators in particular on this, and the number of people really interested in attaining excellence is about as small as the list of certified translators. The “deep learner” translators are actually pretty competent and can do a good job (albeit slowly), but the majority “zero effort, zero knowledge” legal translators basically have zero skill.
The “zero effort, zero knowledge” approach to the translation profession by casual workers can knock the price down to half or even a third of what a skilled professional would offer. However, that comes with a big cost in terms of quality. I have observed hundreds of these types of translators using training software screen sharing over the years, and frankly, they usually don’t know what the text they are translating actually means. If they are translating to English, they cannot perform basic English-language tasks like searching for a document in the related industry and finding a document from the US/UK. As translators, they get employed by offering really low prices and use artificial intelligence to blend together sham translations that look convincing to clients. Since AI can predict what a client will think is good quality quite accurately these days, about 90% of translations these days are just inaccurate sham translations from the zero-knowledge crowd.
A second important issue related to price is effort. A legal translator is technically a professional services worker who is providing translation services by the hour. Generally, in the legal translation industry, the translation rate is simply a proxy for how much time a legal translator will spend on a certain project. A good example of this is a hypothetical Korean Translator. A US law firm could approach them directly and offer a fair rate, or a local employee might be delegated to handle it and offer half the rate to the same translator. What does the Korean Translator do? He cuts the time spent on the project for the local employee in half compared to what the US law firm would get. The difference in regional approaches can explain a lot of why newspapers published in China are written in Chinglish, whereas much smaller papers by overseas Chinese groups published in the United States are written in perfect English.
There is a similar phenomenon in the law firm world. In Shanghai, plenty of big American law firms have a dual pricing model, one for local clients and one for international clients. International clients pay twice as much and get twice the effort, while local clients pay a fixed fee for half as much and get half the effort. Looking at Huawei’s huge international legal troubles and sanctions that cut its revenues in half during a disaster in 2020, and the massive wave of Chinese company SEC de-listings in 2022, I can’t say that the nickel and dime strategy when retaining professional services is paying off at all. The same goes for translation.
