| Renato |
I’m Renato Beninatto. And today on Globally Speaking, Michael has a conversation with a guest who is leading the future of localization, especially as it relates to marketing and engagement. This guest was previously here on Globally Speaking sharing with us the core elements of building a global infrastructure. That was on episode 35. Today, he takes us a step forward and we will learn how we can know that what we’re doing is actually working. Let’s let our guest introduce himself. |
| Daniel |
My name is Daniel Sullivan; I’m the Director of Global Content Readiness at Tableau Software. I’ve been with the company for a little over five years. I came here after a few stints at some of the local tech companies, and came to Tableau at a time when they were just starting to go large in their international expansion efforts. They had a lot of challenges facing them with respect to infrastructure, teams, processes, vendors, how and what to localize, and so we spent about the first three years setting up a team for localization, an infrastructure for localization and processes so that we could start handling more and more of the company’s content. |
| Michael |
It wasn’t exactly from scratch that you build it, but you really came in at a very early point that it was an immature… |
| Daniel |
Right. |
| Michael |
…environment. |
| Daniel |
Right. I often describe it as their translation management system was Microsoft Outlook at the time, and their primary file format was Microsoft Word documents in bilingual format. |
| Michael |
And maybe some Excel. |
| Daniel |
And some Excel as well. |
| Michael |
That’s always like a sort of core element. |
| Daniel |
Yep. |
| Michael |
Last time we talked to you, we focused really on big data, KPIs, where companies can start there, which really led us into this whole idea of, like, setting up an enterprise platform. |
| Daniel |
Right. |
| Michael |
And there was a large operational component to it, in content. It was extremely practical, I think, for our listeners. |
| Daniel |
Right. |
| Michael |
And today, we’re sort of revisiting some very practical things you’re doing more recently. |
| Daniel |
Right, correct. |
| Michael |
Yeah. |
| Daniel |
Some of our biggest challenges that we had was: how much of the website has been localized? and what sort of a state is that in right now? And that’s where, like, a lot of the data really started to come in, too. And fortunately, since I was at a company like Tableau, we had a product that would just sit on top of any data source that I needed, and we could start articulating that story of, like, what state are we really in? Where do we need to make more investments? What do we need to do to get this website—but not just the website, and not even just our corporate communications and how we’re doing that—but just the company as a whole, in its holistic approach to international business and the content required to support that? |
| Michael |
What’s the challenge of understanding how much is translated? I mean, if it’s in WordPress or Drupal or some kind of standard web format, what was involved with that for you? |
| Daniel |
The biggest challenge was just getting access to the data, and fortunately for us, we use an open source platform. We use Drupal here for our website, and they had already set up a lot of data sources from our website data that was available to us, and it was just a matter of looking at and understanding that data and what was required in order for us to start extracting the story of our website from an international standpoint. And so, it’s not just what content’s available, but what user stories does that content align to? So, it’s one thing meaning, you know, how much of the website has been localized? That’s one question. But how much of the website needs to be localized to support those various customer stories and customer journeys that drive our business? And so, it’s not a matter of ‘are we at parity with English?’ It’s a matter of ‘Do we have everything we need to be a functional and efficient business internationally?’ And then, what do we need to tack on to that to support our various other programs, be they marketing, support, legal, whatever they may be? |
| Michael |
Yeah, because parity may not be the answer. |
| Daniel |
Parity is definitely not the answer. I mean, especially for a company that’s, you know, US-based like we are, we’re always going to have much more content available in English, just because the content produced from here is English-centric. The other thing is, we just don’t need a lot of that content internationally—not just because it doesn’t necessarily align to services that we don’t offer internationally, but that maybe some of those topics and those things are just not relevant internationally. |
| Michael |
And before we get too far away from it, can you provide just a simple description of a buyer journey or a profile? |
| Daniel |
Sure. I’ll start from the search standpoint. They may go onto Google or some other search platform, and first of all, they may not even be looking for Tableau because they may not even know we exist yet. They may just be looking for solutions for problems that they have, so data visualization tools may be something that they search for. We want to make sure that we have content that is relevant to those queries that align to our business and align to the discourse that we are a part of, and then pull those customers in to our site and making sure that once they’re there, they’re able to have easy access to any more information they need about what is Tableau, what is our product, what does it do, but then also seamlessly go through a trial process or even a buy process as well. |
| Michael |
So initially, when you start a program, you’re finding out what content is relevant… |
| Daniel |
Right. |
| Michael |
…what needs to be localized, and then if you already have some content localized, is it any good? |
| Daniel |
Right. |
| Michael |
Is there quality, and is what I’m getting now from the new sources quality? |
| Daniel |
Exactly. The first thing we had to focus on was we needed to operationalize how we localize content on our website and how we operationalize content that we use for customer engagement via email and other communication channels. Then we needed to flesh that out with what were the rest of the company’s needs to support that? And so, we went from initially supporting just the website and just email, to also pulling in other products and services that we offer here—say e-learning, training, or certification or our training video library—and started to supplement that with other content that was produced not just within marketing, but by other content producers across the company. |
| Michael |
When you’re first estimating the quality, what sort of data are you getting? |
| Daniel |
So when we were initially setting things up, the other problem that we had was voice and translation quality, and it wasn’t that our vendors were doing a bad job, you know; part of that was just that we had never even really articulated to them our own expectations about how we wanted content to be produced in other languages, how that voice should be defined, what it should sound like, what its formality levels are, and so we started focusing really heavily on things like terminology and style guides, the end goal being that we wanted to make sure that not only did we have the infrastructure for producing content at scale to support our international business, but we were able to produce content that met our quality bar, because the thing that’s going to be affected the most is your brand, and the last thing that you want to do when you’re going big internationally is to say, “Hello Japan, we’re here, and here’s all of our content, and we’re sorry that it’s so subpar, but that’s all you get.” [Laughs] |
| Michael |
“Welcome to our brand.” |
| Daniel |
Right, so I mean, that is a problem in itself, and so we were really focused on quality, and not just finding the right linguists, and not just working with our marketing managers and other stakeholders throughout the company who are going to be consuming and using this content as part of their business functions. About our first three years were focused very exclusively on quality, having an infrastructure for measuring quality, and then monitoring that over time to make sure that the quality was no longer a problem. |
| Michael |
And I just want to highlight one thing here, because I’ve had this conversation a number of times recently. Some people may hear you talk about that brand piece and all of that and they may say, “oh, the preferential stuff.” |
| Daniel |
Right. |
| Michael |
And I was in a meeting where my day job was actually providing LQA services, and they were saying, “Wow, you’re rating this other company really high, and yet our in-country stakeholders have so much feedback and they’re not satisfied.” |
| Daniel |
Right. |
| Michael |
And I said, “Well, we’re months into this; that makes sense,” and they go, “What do you mean?” I said, “Because we’re looking at quality from the perspective of errors per thousand words.” |
| Daniel |
Right. |
| Michael |
And if I say to a linguist, “Hey, we’ve gotten all this feedback,” they go, “Oh, that’s great, it’s preferential.” There is an evolution there that… |
| Daniel |
Right. |
| Michael |
…goes from errors per thousand words to actually meeting expectation. |
| Daniel |
Exactly. A marketing manager may come back to me and say, “This is not good,” and it’s not for me to come back and say, “Yes, but its error rate per thousand words is this, and so that means it’s actually okay.” |
| Michael |
It’s like, so what? |
| Daniel |
Right, so what? Exactly. And so, for us, the quality KPIs that we capture, and we’re capturing now, that’s not the final adjudicator on whether or not this content is meeting our quality bar anymore. |
| Michael |
Yep, because you guys have done something interesting, because it’s not just dependent on that person giving you feedback. |
| Daniel |
Right. |
| Michael |
It doesn’t end at, like, sort of preferential. |
| Daniel |
Exactly, and this I think is where our editorial program really starts to shine, because we do have an extensive pool of editors that we work with. A lot of them have been with us for over four years now. They have looked at every single piece of content that has come through our system. They’ve watched every single training video for multiple versions of Tableau going all the way back to 8.3. They’ve read all of our legal documents; they’ve read a lot of our job descriptions; they’ve read all of our training manuals. I mean, they’ve seen everything, and so they know so much about our brand, the next step there was to start inserting some of them into our local offices. And so we’ve been doing a lot of testing with that, and one of the places we’ve had the most success is in France, where our editor actually sits with the marketing managers, sits amongst sales, gets all of this feedback from them and through her, we’re able to start to construct what that Tableau voice is, how it should sound, and make sure not just that it’s aligning to, say, a term database or what we have in the translation memory, or how we’ve done things before, but make sure that it’s really aligned to that discourse that we’re trying to speak to. Because ultimately Tableau does sit within a distinct discourse—and again, that is about data visualization, data analytics and things—and there is within that discourse already a set language about how and what people talk about. And just to give you an example: if you’re talking to a carpenter and you say the word ‘hammer,’ to them, that’s going to mean one thing, whereas if you talk to a geologist and you say ‘hammer,’ that signifier is going to signify something completely different, right? And so, within our discourse, we need to make sure that we are aligning to what people are looking for, what people are asking and who our real customers are, and making sure that we’re aligning to that as much as possible. |
| Michael |
That’s really opening up communication channels; it’s really customer-focused. But what are you moving towards as the ultimate test for quality? |
| Daniel |
We spent the first three years building an infrastructure and focusing on quality, and once we had gotten to a state where I felt that we had the data that we needed to be able to report on our quality metrics, we had the infrastructure to produce multilingual content at scale, and now the next problem was how well is that content performing? So, is it actually doing what we want it to? And I think this is where we start to diverge from what I would call traditional localization processes. We start to become, I think, more of an out-of-box thinker with respects to how we go about doing this. And so, about two, two and a half years ago is when I first started approaching leadership, because at the time, we didn’t really have a robust SEO strategy around organic search, and so I approached leadership and proposed to them a program where I saw SEO being a component of localization. Because localization, as you know, is one of these big investments that companies make when they go global, and one of the challenges with it is always going to be justifying that spend. So, what is it actually doing? And it’s a perennial problem in this industry where we talk about ‘what is the ROI of localization?’ and this just keeps coming back. |
| Michael |
Yes how do we make it not a cost, but a revenue driver? |
| Daniel |
Exactly! And so initially, I started thinking about that and I thought ‘no, no, you’re going about this backwards; it’s not about how do I make this something that has ROI in it, but we need to take it further upstream from there.’ And so, it was more about ‘how do I take what we’re doing now and cut out a lot of the fat and a lot of the steps necessary so that we’re producing not only quality content, but it’s also being optimized just as a part of that regular translation process?’ So, we started developing and thinking about ways we can produce, what I would call it, optimized localization workflows, and just bake into the process a lot of the intelligence, a lot of things that go into producing content. Not for SEO purposes, but producing content that is aligned to what users and potential customers within our industry, and again, within this discursive space, what are they actually looking for? What are the problems that they’re having? And can we start aligning more to that and producing content that not only leverages everything that we’ve done to ensure those quality bars are met and that it’s aligned to brand voice, but that the content we’re producing also is fulfilling a business function for us? |
| Michael |
Many of our listeners may have heard of SEO, they may be involved with SEO, but search engine optimization, why is that such a key? What’s the brief description of what it is and how that begins to feed this entire global content readiness process for you? |
| Daniel |
There is a subset of users out there who have, within our discursive space, a problem, and they go to a search engine of their choice and they start just searching for answers to their problem. And they type in various strings in order to get search results that they hope align to what they’re actually looking for, and ideally something that comes up in those search engine results is going to give them the answer that they’re actually looking for. And what Google does is it starts monitoring questions people are asking, and those questions are actually what you call keywords. And a keyword isn’t just one word; it can be multiple words. It can be a whole string; it could be a question. And so, it’s monitoring what most people are asking, and then what content, just based on what people are doing and their behaviors, is actually providing the most relevant answers for them. And I latched onto this because while we were producing content for a lot of the marketing efforts we have, be they paid or direct or whatever they be, we produce content for these campaigns. It was one thing for us to hand off to our stakeholders ’here’s this beautifully well-translated content, well-constructed multilingual content, ready to go for your campaigns.’ That’s one component of what that’s supposed to do. But then irrespective of that, irrespective of the various programs or whatever marketers are doing to get that content to help them drive the leads that they need, I wanted that content to also start aligning to what people were searching for so that it had an organic component to it. I am not a marketer; my team are not marketers. But the content we’re producing I wanted to make sure not only supported those marketing efforts, but was also producing this other function that didn’t require hands-on marketers to help drive, and that’s where the organic search component comes into it. So again, this content’s working great for this campaign, most of which is being driven by our marketing efforts, but it also has this other function; it’s pulling in traffic organically which has nothing to do with what we’re doing other than how we’re constructing content and the words we’re using to talk about our product, our business, and how we’re putting that on the page. And that I can have an effect on, and that my team can have an effect on, and that’s what we’re focusing on now. |
| Michael |
Yeah and how you’re creating the content. So, you’re localizing it? |
| Daniel |
Some of it we’re localizing, so we actually have about three different processes we use. One of them is looking at our website more holistically and looking at it from what I would call a keyword grouping or a keyword matrix. For example, one of those might be ‘analytics,’ and then within the keyword group or keyword matrix ‘analytics,’ there may be different types of analytics, and so producing content around those various types of analytics and analytics questions that people have. Finding on our site what’s already available, determining whether or not it’s optimized already or if it has opportunity to perform better, and then sending those packages of content out through a translation and optimization workflow brings that content up to speed as a holistic experience. And so, what you’re trying to do is capture content. Let’s say you’ve got a couple of really good blog articles over here on consumer analytics, and then you’ve got a whitepaper on consumer analytics, then you might have a customer story or two about consumer analytics, and they’re all aligned in this keyword matrix, but it’s only been produced for quality reasons; it’s never been produced for whether or not it’s optimized for search or not. And so, we’ll go out and look for those things, grab those groups of content, run them through an optimized process, put them back on the site and then start monitoring them at 30-60-90-day intervals. The other two processes: one is what I call optimized localization. We’ll produce a net-new piece of content, we’ll optimize it on the front end, we’re designing it already to align to a strategically important keyword that we’re aware of, optimize that content, build it correctly for search from the start, and then run it through a translation workflow which includes an editorial step that also includes those optimization checks as part of the process, so that when the content is finally published and put out there, it’s already been through a lot of these really important steps for SEO before it’s even been published. And so, it’s pre-optimized and we start watching and monitoring that to see how well it’s performing. The other workflow doesn’t involve translation at all; it’s identifying keywords that may not have an analogy in English or in other languages. It can’t go through a translation workflow; it needs to be built net-new because it’s a keyword that’s specific to Germany, let’s say. This is where it becomes really crucial for my team as a whole and why it’s important that the same resources that are net-new building that content as a copywriter are also a part of this translation workflow and this optimized localization workflow that I described. Because there’s so much intelligence that’s created with every iteration that we run, and I want to make sure that all that intelligence is baked into a single team that knows the content we’ve produced, they know what we’ve done in the past, they know the choices we’ve made in the past, they know what works and what doesn’t, and so it just makes it so much more efficient than if we had, say, an entirely separate team of copywriters for Japanese, for example, who was doing their own thing over here, and another team over here who is part of the optimized localization workflow and they’re separate now. |
| Michael |
So, it kind of builds on top of each step. So, the editor is standing on the content that the translators have done that has been optimized, but then the editor is going out and creating something totally new which then is being fed back, which then can become a part of the regular work of the translators. |
| Daniel |
Exactly, and I’d be remiss if I also didn’t include here that partnership that those editors have with our regional marketing stakeholders as well; that’s really crucial, too. It’s not just a check and balance; it also is really great for our marketers because they know exactly what is being produced and what it’s doing. |
| Michael |
Does this work that the editors are doing—they’re getting ahead of the curve in-market in some ways—is that something that’s present for almost every company? Does it have something to do with being a nascent tech company, that there’s sort of new terminology and new thinking? |
| Daniel |
No, I think you’re going to find that about anywhere. What’s unique for us is that since SEO came over to my team so early, we’ve really made sure that that was one business function that didn’t have a strong US-first component to it. And so, from the moment we started building out our strategy, it was international, and it had that multilingual component built into it. I don’t think a lot of companies are doing what I’m describing to you right now; they may be doing it in some fashion, but I think it’s usually that the teams are a little bit more disparate. That puts us in a different bucket, I think, then, say, a lot of the other traditional enterprises who have a very traditional operational procedure. Just by the nature of our business and who we are in our age and where we fit with other technology companies, we’re basically forced to do a lot of out-of-box thinking; we’re forced to do things different. And so, for me, the biggest flattery you can give me is that I’m not doing things conventionally. I mean, that’s great! I don’t want to do things conventionally. I want to do things differently. And I really want to make sure that we are moving the needle and we’re doing it in ways that other companies aren’t, because we’re ahead of them or we’re thinking differently or we’re just really strong out-of-the-box thinkers, and we’re really open to taking these risks and trying new things, and that’s what I want to be. |
| Michael |
Yeah and this has effects on some of the more traditional processes because you’re being non-conventional. Talk about something like terminology and then maybe translation management. |
| Daniel |
Terminology is really important to us because of the search component that we have. It really is for me, I think, a forcing function to ensure that the ‘what and how we’re talking about Tableau’ is aligned to the discourse that we’re already a part of. And I’ve already described to you about the difference between, say, the carpenter and the geologist and stuff, but it makes sure that the way we’re talking about Tableau and our products is already aligned to a discourse that was there before Tableau existed, and so that ‘what and how we call things’ is already aligned to user expectations. |
| Michael |
You mentioned an instance with naming conventions. |
| Daniel |
We’ve always had a product called Tableau Reader which is similar to Adobe Acrobat Reader. It’s a free product that we have that allows anyone with the product to view and read Tableau workbooks. Last year, we started making our transition to subscription and as a component of that, we came up with license profiles, one of which was called Tableau Viewer. And this decision was made without any knowledge that, before we had even come up with this SKU, Google was already servicing search results when people did a search for ‘Tableau Viewer,’ and it had figured out that when people searched for ‘Tableau Viewer,’ they actually meant ‘Tableau Reader.’ And so, when someone searched for ‘Tableau Viewer,’ it sent them to our Tableau Reader page and our Tableau download page. Last year, when we made our transition to subscription and added a new SKU for Tableau Viewer, when people were doing searches for that, instead of being sent to our purchase page or sent to our, you know, details about the Tableau Viewer license, they were still being sent to ‘Viewer.’ That was something that really highlighted the power and implications of search, and how, before you do start naming things like that, search needs to be a thing that you look at, because you can have downstream ramifications of a decision like that that could be expensive to resolve. Now fortunately, we have a crack team here, and once we had identified the problem we were able to resolve it, but that just highlights right now the importance of terminology and naming and how we go about things, and that search and what is already a convention out there need to be something that you take into consideration when you’re naming things like that. |
| Michael |
What about translation memory? Are you still using that in the traditional ways? |
| Daniel |
We’re still using translation memory in traditional ways, and we’re also starting to not use it in other ways. I mean, we’re starting to feel that there’s less of a need to have a reliance on that, and especially for the content that we’re doing. It’s our gold, you know, that’s our data, but there are some areas where, when your concerns about preserving the translation memory start to negatively actually impact your time to market, it can also start impacting your efficiency, and I’m thinking really specifically about video here. You know, we’ve always run all of our subtitles through a translation workflow because of that translation memory thing, and because we wanted to make sure that it was really easy to repeat and if there was a delta in something, that it was just really to expose just the delta. You know, these traditional concerns here. And now, however, with all the types of new technologies that are coming out that make video localization so much easier, I really don’t care about the translation memory as much as making sure that that video is produced in a way that resonates locally with our customers and prospects and is aligned to our messaging and our branding and our voice and our terminology. But if I don’t have access to the translation memory for that content…it’s so unique that the chances that you’re going to leverage it for anything else are very, very low, and so, it’s more important to me that our ability to scale and our ability to turn around that content much faster is more important than preserving the translation memory for it. |
| Michael |
That’s a really good thing to hear and for folks to be aware of and in conversation about. Like, are you getting the benefits that you are perceiving? |
| Daniel |
I would also add to that email as well. Email is just not conducive to translation, and we know this from data because we’ve actually tested this. I have a German editor on my team, and it was early in our program where we were producing these translated emails trying to advertise for one of their webinars or something that they were running. And they weren’t getting the response that they thought. You know, it was like, “why are we not getting the registration levels that we anticipated?” And so what Yannis did was like, “You know what? I’m just going to rewrite this thing from scratch as if I’m targeting that audience and see what it does.” And you know what? Boom—the moment we did that, the registrations went through the roof. And again, those emails, they’re so topical in nature, they’re so point-in-time; they have such a very small shelf life that for us to obsess over preserving the translation memory in cases like that, I think the business impact of it is much more important. And also, you know, that kind of content can be produced pretty quickly on its own, and I think translation memory actually just gets in the way there. |
| Michael |
Yeah. Well Daniel, we’ve covered a lot. I think one of my big takeaways, and you started early on with this, is that it’s listening to the questions that your customers are asking. |
| Daniel |
Yes. |
| Michael |
So that immediately makes search a key part of any successful performance-based program. |
| Daniel |
Yes. |
| Michael |
And I think our listeners, if they’re looking for a place to start, that would be a good one. What would you recommend that they do to get some ideas around what their customers are asking for? |
| Daniel |
Any company nowadays is going to have a team that’s dedicated. If they don’t have a team dedicated to SEO and organic search, they definitely have a team dedicated to paid search, and there’s a ton of knowledge that they have. And it’s really easy for them to not only test, but start to extrapolate from very focused, locally targeted campaigns what people in certain regions of the world are looking for. And I would just start with conversations there, learning about the process, how it all works, and find out if there’s a way that you can start baking this into what and how you’re doing things. 2018 was the first year we really operationalized this and put it into place, and the impact we had on the business is quite phenomenal, and I have all the data to show that. The localization ROI was less of a concern; it was more about how can we make this content that we’re producing have a real driving business function? And we have data and we know that it does. |
