Fake news in the localization industry

What does it say about the localization industry when key players in the industry feel the need to apply rather loose analytical criteria in reaching clearly self-serving conclusions?

As I’ve stated in a previous post, it is disappointing to see that many in the localization industry equate globalization with technology-enabled translation. In the drive to keep pace with the rapid proliferation of internet content, larger Language Service Providers (LSPs) have made important technology investments in machine translation (MT), content management system (CMS) connectors, APIs and the like. It is perfectly logical to publicize the value of these technologies, for indeed they certainly can be implemented to assist globalizing companies in more effectively managing their multilingual output. However, in the past few weeks I’ve seen three examples of important industry voices offring bold but empty claims that implicitly make a case for technology investment, using language that obfuscates or conveniently repositions what the underlying data actually say. This sort of marketing does not honor the real value of localization industry tech. Instead, it attempts to support the status quo by any means necessary, diverting the corporate buyer’s attention away from more holistic approaches to globalizations by keeping the focus solely on technology.

What does it say about the localization industry when key players in the industry feel the need to apply rather loose analytical criteria in reaching clearly self-serving conclusions? Kirti Vashee (@kvashee) wrote an interesting article earlier this year about a similar phenomenon in reviewing the Lilt study comparing MT quality. Vashee states, “I think it might not be over-reaching to say that this effort is neither fair nor balanced, except in the way that Fox News is. At the very least, we have gross self-interest pretending to be in the public interest.”

In the example I discuss below we see more of the same – entrenched self-interest masquerading as industry best practice and inspired prescience.

Hard data? Or misleading conclusions?

Common Sense Advisory (@CSA_Research) has long been a powerful and valuable voice in the industry. I’ve subscribed to their advisory reports and have found them to be useful resources. That said, in a recent post, it was clear that CSA seems willing to tailor its conclusions to say what the industry wants to hear. Preaching to the choir may suit CSA’s bottom line, but it certainly does not constitute the kind of rigor one would expect from the industry’s leading source of market analysis. Moreover, they have missed an opportunity to bring to light important trends that are changing the way consumers perceive quality and what kind of future we are moving toward.

Using the tags “Best Practices” and “Business Globalization,” CSA put out a teaser blog post to publicize a recent report designed to arm industry professional with data-driven justifications for capturing more localization budget. First off, the title “Hard Data to Support Global Growth” is misleading. While I do not have access to the full article, the blog post offers no clear mapping of localization investment to better global results. CSA allude to potential growth drivers by pointing out that each full-time localization employee “drives” $2.5 billion in international revenue. However, it seems ludicrous to me that CSA would expect a client-side localization manager to use that nugget of data to convince executive management that her team should be allotted more budget. It should be completely obvious to any observer, localizers and laymen alike, that the $2.5B is not a direct outcome of having localization employees. This is a flagrantly misleading and a useless statistic. To calculate ROI there has be to some causation at play. While a localization employee certainly plays a role in ensuring that global roll-outs happen successfully, there are numerous other actors in the corporations and types of investment that will be necessary to deliver international revenue. That CSA would use such an oversimplified set of calculations and then use verbs like “drive” to exaggerate the impact a single localization employee has on international revenue makes one wonder whether CSA really believes that its customers lack the common sense necessary to see through such a hollow conclusion.

CSA Hard Data Image

Adding insult to injury, under the subhead, “Sad News for Perfectionists: Some People Prefer Bad Translation to None,” CSA takes a swipe at those in the industry who still value nuanced quality. The idea that “good-enough” is good enough has long been an argument by those who favor MT even when it is not called a strong tactic. Not surprisingly, it is often those with the most to win through MT (MT vendors and larger LSPs with amortizable investments in MT) who make this argument.

The CSA graphic attempts to illustrate that “most respondents” prefer low-quality translation to English. On closer examination one sees that the majority they speak of is a meager 51% majority, a figure that I imagine is within the margin of error for the study (which conveniently has been omitted). Aside from exaggerating this claim, the analysis misses a much more powerful trend. With each successively younger generation that CSA has polled, there is a clear move away from the acceptance of the “something-is-better-than-nothing” content philosophy. Among millennials, the emerging power demographic both by virtue of sheer numbers but more importantly due to the influence they exert in the consumer market, the majority of respondents in fact indicate that they disagree with the assertion that poor translation is better than nothing. As millennials continue to demand greater brand authenticity and unabashedly pick winners and losers in our age-of-the-consumer economy, how has CSA neglected to point out this rather obvious and important trend? The only rationale that seems to make sense is that CSA does not want to alienate some segment of its supporters by suggesting that there is even a remote chance that human beings cannot be suckered into accepting subpar language quality, delivered by mediocre MT engines, even when their own graphs illustrate the contrary.

“The sky is falling, the sky is falling.”

Although originally published mid-2016, Grant Straker’s article on LinkedIn portending the apocalyptic demise of a third of the companies in the localization industry saw a flurry of activity recently. Mr. Straker (@gstraker), CEO of Straker Translations, bases his menacing projection on three causes: some new movement of content online (not sure I even get this argument, given that dealing with online content has been the bread and butter of the industry – small and large LSPs alike – for quite some time now), easier vendor comparison (making rather broad assumptions that all buyers will shop for the lowest price, period) and MT creating winners and losers when it comes to margins. Ironically, Mr. Straker fears that translators will abandon small LSPs because in order to maintain profitability these losers will try to squeeze talent by demanding lower rates. Ironically, my conversations with translators has revealed quite the contrary. Many seem to prefer working with smaller LSPs precisely BECAUSE they tend not to put translators through the wringer.

But I digress. Let’s dig into the numbers to see how realistic the claims are, root causes aside. Mr. Straker points to some Common Sense Advisory (CSA) numbers in setting the stage for his analysis. He indicated:

“There are an amazing 18,000 translation companies in 137 countries in the US$38 billion translation industry according to leading research group Common Sense Advisory. 60% have between two and five staff and with average revenue of $1.4m most companies are relatively small.”

Assuming these are accurate metrics, let’s examine Mr. Straker’s bold claim mathematically. If one third of the 18,000 companies worldwide were to disappear, 6,000 companies would either close shop or would be acquired through M&A activity. Assuming that the industry as a whole is growing and not shrinking, an assertion that CSA’s annual reports clearly supports, a grand total of US$8.4B would need to be absorbed by existing or new language services providers (LSPs), if we are to believe in Mr. Straker’s prescience.

Having been a scrappy, small LSP owner myself I can assure you that these 6,000 companies are not going to die out without a fight. Although small LSPs may be weaker financially and technologically than the larger companies in the space, their smaller size makes these LSPs significantly more agile and able to adjust to emerging market conditions. This adaptive advantage is particularly strong when compared to the LSPs who are heavily focused on amortizing their technology investments, betting on a fully automated language services world of the future.

Given that smaller LSPs may be relatively resilient, M&A activity would need to absorb a rather significant chunk of the US$8.4B in revenue. How realistic is that? Given that Lionbridge, the largest company in the industry, is barely a US$500M company (a mere 1.3% of the industry total), I fail to see how M&A activity could accelerate so dramatically that 20% of the throughput for the entire industry could be redistributed among new acquirers. Even if each of the top 100 firms globally were to somehow find equity partners to finance these deals (a nearly impossible order to fill considering the competition for equity and venture dollars in today’s economy), each one would have to acquire (and, perhaps more importantly, integrate) 60 companies a piece, consolidating US$84M of economic activity, across multiple geographies – all within 3.5 years!! Given that a number of the companies listed toward the end of the top 100 list have revenue in the US$8-12M range, it seems unlikely that they will have the wherewithal or, indeed even the motivation, to leverage and pull-off these kinds of deals.

Needless to say, it’s rather disappointing to read the comments of Mr. Straker’s post and observe so many in the industry who are eager to rather blindly to support what seems to be a well-designed publicity stunt. If one looks through the comments on Mr. Straker’s post, one sees many who refute the claims with smart logic and only a few (suspiciously, several come from those who work at MT companies) supporting Mr. Straker’s vision for the futre. That said, it’s easy to see why this sort of fear-mongering if useful to a larger LSP that is actively luring small LSPs into selling out for the lowest possible price.

Furthermore, the assertion that smaller companies are not well positioned to differentiate themselves is just bogus. Increasingly, one finds firms in the sector who focus on particular niches (language services for search and discoverability, marketing adaptation, e-learning, multimedia, embedded services, and many more) and finding great success. Now, if the argument is that small companies won’t be able to compete with the localization firms that decide to convert their businesses in the MT-driven word farms, then I agree. But small firms already know that they are not going to win that battle.

I believe a converse argument could be made. If we look to the future (and by future I mean long term, not a ridiculous 3.5 year timeframe) I can imagine the industry reaching a point where firms that focus too heavily on MT and hyper-automation may find that they will have squeezed themselves out of the industry.

It might be best to illustrate this with a metaphor I was discussing recently with Kirti Vashee. Let’s imaging that supplying translation is like moving water from the well to homes where it is used. In the early days, all of the different water vendors compete over who has the shiniest buckets, who has the purest water, who can get the water to you fastest. Over time, the vendors who invest in automating the movement of water to homes begin to see great success. People love getting their water faster and cheaper. But pretty soon a new player comes into the market and builds underground pipes to get the water directly from the wells to the homes, benefitting from the work the early investors did to automate and speed the bucket delivery system. With the pipes now in place the water is just a commodity and the early tech investors are out of business.

And what remains? Well, what is left are the vendors who are thinking about what you actually do with the water once you can access it. The smart vendor is the one who is thinking about purpose and application, linguistically and culturally, and not just creating the fastest and cheapest pipeline. I would assert that many of the smaller LSPs are in fact the ones who are focused on nuance, purpose and specialization. Far from seeing their prospects shrink, I believe it is they who will see a plethora of opportunities arise and who will have the agility to adapt to the new frameworks.

What about the water pipes, you ask? Who will own those – Welocalize, Microsoft, Google? Let’s put that question to people like Kirti who may have more clues as to where the future may lead.




  1. Great article.

    I agree that the companies that focus too heavily on building MT fast will run out of business.

    Even if the MT software is improving all the time you still need to keep them updated with high-quality translations. And as most LSPs rely on freelancers and still tries to find the cheapest solutions, they will just build MT engines that are not really worth a lot. You might double the through-put per person, but so far the quality is still lacking a lot at that speed. That might improve time-to-market but at a lower quality.

    Just a few years ago you needed millions of words of quality content to build a decent MT, but today you can actually train an MT engine to a decent level with a lot fewer words. So from a business perspective smaller agencies or even freelancers will be able to build a useful MT engine. As a freelancer that opens up the opportunities of accepting still bigger projects, and you will be able to produce high-quality translations faster at a lower cost.

    I think that the LSPs forget that their main value is actually their network of freelancers. Some of them would love to get rid of that expensive part of their business, but smaller agencies and freelancers will actually figure out that they can offer the end-customers more at a lower rate than the LSPs can.

    On top of that I think that some LSPs have lost track of their main purpose: delivering translations. I see more and more projects where the end-customer chooses a different LSP because they are unhappy about the quality. 5 years ago the end-customers were shopping for the best rates, but more and more companies are now shopping for quality. But if you ask which things we need to improve the LSPs never really knows, and they still base all payments on the TMs or MT engines that contains the translation quality that the end-customer didn’t like.



Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: