S3 #07 Catering to the widest possible global audience

[00:00:00] KRISTINA PODNAR, host: Artificial intelligence and localization were made for each other. For the first time in history, we have the power to make localization more authentic through AI. The thing that's interesting to me is where does marketing play a role?

[00:00:12] INTRO: Welcome to The Power of Digital Policy, a show that helps digital marketers, online communications directors, and others throughout the organization balance out risks and opportunities created by using digital channels. Here's your host, Kristina Podnar.

[00:00:27] KRISTINA: Hila Shitrit Nissim is a marketing veteran and the chief marketing officer at Blend, an AI localization platform that works with brands globally to help implement stronger localization practices and strategies. What's interesting to me about Hila is that she has over 20 years of experience in marketing. And I can't imagine a better person to have with us today and ask every marketing question we want to know and a few things as well about AI and localization. Hila, welcome.

[00:00:57] HILA SHITRIT NISSIM, guest: Hello! Thank you.

[00:01:00] KRISTINA: Well, if you can't tell, I'm jumping out of my seat; I'm really excited about diving in with you today because, you know, you have such a wealth of knowledge, not just about localization and AI experience. That's the space that you're focused on now, but you have this long career of marketing. And so, I think that you bring a unique viewpoint, because it's, it's a very rich perspective. So why don't we take a step back first and ask you to explain localization versus AI localization? And is it one versus the other, or based on your marketing experience, what does the ideal solution look like? Or is there an ideal solution?

[00:01:38] HILA: So, let me try to explain localization generally, the process you do in order to transform ideas, content, businesses into new different markets, new different languages, cultures. And so, there are many ways to go. And in the past, everyone was talking about translation; now the localization is the bigger concept because there is way more to it than just translation, many aspects, many types of content, mediums, and nuances. And from the content itself through the market research, SEO keywords, and even product customer experience. So, there are so many things to think about when you are entering any new market or serving audience in a different language. So, there are many ways to do a localization, from human experts or linguists they are doing the work from scripts, and, there are AI and machines that help with the process. The way I see you cannot do just one or the other. I think obviously the human, the human option is right. It's way more expensive, complex, and time-consuming. And that's why we are using today lots of AI technologies machine learning and to make the process more effective faster, save some money and help the human translators, the human linguistic experts do their jobs better. So, we start at Blend, start with the machine as much as we can. We try to process the content with AI technology. And then we are bringing the experts from different locals, from different parts of the world to verify, to make sure the machine got it right. At the end of the day, you cannot make without the human touch or human aspect. And so, I think it's a combination. I think today, most companies in localization, together like AI and human parts.

[00:03:54] KRISTINA: And you mentioned machine learning. So is the role of the translator different? Are they there to train sort of the AI or what's really the role of the translator today?

[00:04:05] HILA: Right. So, it's a good question. We are training machines every day. So, whenever a content piece is going through this process of AI processing by the AI machine, and then it's being delivered to the human experts to really, to review every correction the translator is doing, going back to the machine for next time, the output will be better. So in a way, the human translator is helping us to train the machine, but his role is really to make sure to know that the machine did not do wrong but it's right. No mistakes.

[00:04:47] KRISTINA: I see that a lot when I use the Google translator; my native language is Croatian. And so, I like to look at the translations from English to Croatian or Croatian to English. And if I ever need a pick me up, I go to Google and look at their translation because I think of Google as the holy grail of translation of machine learning of AI. And I realized that we're not even like walking; we're just crawling. It seems these days. Is that true? Are we just crawling in terms of machine learning and AI for localization?

[00:05:15] HILA: I think it's just complicated, and you hope we always need human intelligence. We must be smarter for the next time. But again, today, there is always this human patch that we must add to today. Whenever it's the content is the customer-facing department, the content needs to be an operating part. It's not for internal use. If you're going to Google translate sometimes to understand if a friend had posted, the only thing that's fine, you got it right. 95% right, 85% right, but when the business needs to localize, you know, it's product or website, you have to have 99.9% right. That's why we cannot take the risk of trusting the machine solely. So our community of translators, so we cannot do it without them. It's a big part of the process. And we tried to put as much as we could into the machine, but the output must go through this quality assurance and everything.

[00:06:25] KRISTINA: Is that always going to be the case because I'm thinking about where we are today. And we certainly are in uncertain times, right. Brands are having to adapt and succeed not just in a globalized world but in a very polarized one. So how does a machine understand that? Can it, or is it down to the human then?

[00:06:43] HILA: It's not just translation. So, it's not just saying the same sentence in a different language. In many, many cases, you have to say differently what we call transcreation or even content creation. You have a certain landing page or campaign English for the U.S. market. But for Israel, it probably be totally different, you know, grammar or language, because Hebrew is so different than English in terms of how you treat female and male, like so many verbs are, so sometimes you have to have human attention in order to convey the same message, same message, but that it would sound like native look-alike. Not something that was written in another language that was translated. So, I think as long as businesses and brands maintain this feeling or concept of talking to people as, as people and becoming natively, local. They need this local touch of an expert that lives in the same country; when it comes to translation, I think we might get too high, high a percentage of accuracy and maybe this will, I don't know, be done by the machine, completely in the future. But I think when it comes to really communicate and talking to people, I think transcreation content creation in different languages will still need to be able to do the job. Although today, there is also an I love this for Content creation. Right? So, we are getting their ideas.

[00:08:27] KRISTINA: So that's a great question because you've said previously that content is king. Are we at a point where machines can be trusted to create content into localized content for us? And how much of that freedom should we give to machines because we might want to experiment, but how do we balance the opportunity and the risk? Because if you make a blunder, it could cost you!

[00:08:50] HILA: I don't know it's beyond my ability to forecast, but then I guess that's where the world is going with, you know, it's synthetic voice today and AI text to voice and synthetic video and genetic even like imitating. You create characters; you create images of people that do not really exist today with technology. So, we may be beyond my ability to understand where we get to know where they got us, but yet technology is fascinating.

[00:09:27] KRISTINA: Where do you see the rule for that technology and marketing? If marketers want to experiment, whether it's with localization and using AI localization versus human localization or using even synthetic data, from a purely marketing perspective, where is the role for that experimentation? How should people experiment with that before they just unleash the capabilities globally?

[00:09:51] HILA: Yeah, I guess there, there, there are many ways through experiments today. When you want to start localizing or penetrating a new market, I think it's a bit easier than in the past, where you need to go and find translators, screen them, and hire them and educate them on so many different files and systems. There are like very easy-to-use tools and technologies that make the process easier, make it faster, and even, you know, less expensive. So you can experiment with the lower budgets and see how the market responds and how the traffic is behaving. What visitors say, behave on your website. So I think it makes it easier to start and try it out. Obviously, when you see it works, so you scale, but then that's also part of the advantage here; you can scale very quickly, and you're not limited. So even when it comes to people, our communities, so large we can scale projects, you know, very fast and technology is unlimited, the in capabilities and when it comes to, I guess, video and imaging again, it's very, you know, makes it more accessible, I guess. And I don't think I've used the video synthetic video ad, but I still a very, very nice your things in for you need to scale or create a large number of videos, explainer videos, and saved videos with lots of characters, lots of different languages. So, it's very easy to do it today with certain technologies. I think it's great for marketing because this is what we do. We create content all day. So it makes our lives a bit easier. And, um, yeah. Also, content editing tools that make our language better, you know, so.

[00:11:41] KRISTINA: What about the cost aspect? Because one of the things that I think many people wonder about is the quality versus the cost aspect. There's always this trade-off; it seems I'm leaving the human, which is seen as the ultimate quality assurance if you will. So, it's a high cost and using the human translation localization aspect if we moved to the AI, but of course, you get economies of scale, I would hope. But where does cost come down? Are you finding that there's a perfect balance or a perfect formula for a combination of human and AI-driven localization and translation, or how do we understand the right percentage or the right formula to get the marketing strategy delivered?

[00:12:21] HILA: Yeah, I think most brands do not stress solely on the machine, and the fact that they can combine the two, means that the translators work less time or he has to translate fewer words. So, the payment is cheaper, and he can, or she can do more projects at the same time and make even more money for living. So, I think this balance of working with the localization platform that combines the two, I think it's way more cost-effective, and you are not compromising quality because that's why you want someone that, you know, most of our brands know in person, you know, that the translators they work with, they're part of their team. They have weekly calls with them. They make sure they know the tone of voice, the brand personality, and that's how they are. They, ensure that the content speaks their language, all these languages. And so, I think in these regards, then lots of advantages and also very something that is called translation memory. This memory is something that we build over time, and again, it makes the work less complex or less work for the translator because this memory, you probably are repeating some that terminology some of the sentence of the professional language and then we have these ways to find the message in the text. And sometimes it's even involved 100% because not only the paragraph is the same, but the paragraph before and after are the same. So, you're quite confident that these texts have already been translated and was done, you know, in a good way. And then the human can work less time or less hard. So, all these translation and memory and machine translation, or they really bring lots of advantages in terms of cost time and quality.

[00:14:23] KRISTINA: And so, there's a lot of opportunities there, but I'm also thinking about the challenges because what we're really talking about is a need for a lot of data, a lot of information, and a lot of data. And yet we're living in this crazy world where there are more and more data privacy regulations. And so, at the same time that we need more and more data, we see almost more and more restrictions on that data. How do you advise marketers to start to balance that? Especially when we're talking about localization and translation. I mean, it's a little bit of Pandora's box, I think. So, let's break it down to maybe like from a marketing perspective, just thinking about it from a traditional marketing perspective, you want to be able to connect with your users and to do that. You must be able to personalize things potentially. And the same holds true, probably for translation and localization, you want to be able to localize to not just maybe a group of people, but maybe you want to look at to the individual potentially. And so that requires data, but then if you have things like GDPR, for example, it does put a little bit of a damper on how much data you can have. And I'm wondering does that it makes that impact. And if so, how does that impact the machine learning that we're doing for translation? Or maybe it doesn't; maybe we strip the personal data out. I'm not sure how that works.

[00:15:41] HILA: I think it's the same challenge you have when you are personalizing your email; localization works hand-in-hand with these platforms. So, if you are able to send them there, "Dear Kristina", then all we need to do is understand that is your name and not translate this word, right? Because it's a name. You don't want to localize the name but other than that, the same privacy and personalization standards that the organization or marketing department already works. That's what we connect. For some of our clients, we are localizing their customer support tickets. Let's get a ticket from a client in Chinese or in or in Portuguese and the agent that is providing the services in English speakers. So we are actually getting the ticket, the answering the in the native language of the person and we are localized in Via machine and a person and it's going to send it back to the company and being sent back to the client. So, but again, the date we do not own the data. It's the company's content. We are doing things, translation memory for the client, but it all meets all privacy and security requirements. So when these regards to. I don't see any additional risks, the existing risks that they are today, they time personalization, you know, advertising in general also.

[00:16:54] KRISTINA: Yeah, that's interesting. It introduces interesting requirements between who's a control or who's the processor in the traditional data privacy realm. And so something I think for, for organizations and for our listeners to be thinking about, for sure. I'm curious; I think about small companies because we've been talking a lot about larger companies potentially, but what strategies and tips do you have for successful startups or small business marketers who are trying to really take off on the global market stage at this moment, because they're a unique audience and certainly one that you've had a lot of experience with.

[00:17:28] HILA: So again, it's really, I don't mean you should localize it as a first step, you should start, you know, with the go-to-market strategy, find the product market speeds, find the right audience and then scale and grow. And then, if this requires speaking to other audiences in another place, so obviously, localization is a great way to do it. I don't think things are different from large brands. Actually, I think small startups when they grow, and they see the potential globalization or, you know, many e-commerce, gaming companies, e-learning but also, I don't think companies that find that there is a market for them in China, they need to speak Chinese. So most probably, they will hire a person or two on the ground, but then they will do localization or some kind of a vendor. And so I think it really easy that you need to start and scale gradually. And when you see the results, when you see the ROI, when you see that revenue is swinging, then comes the appetite, and then you keep expanding. And it's good because he can do it gradually. You don't need to start one day with 30 languages. Most companies do it one after the other. And sometimes, you know, they can be, they can see a huge success in one market failure and the other, and then maybe has to do with the product, with the competition with the audience.

[00:19:04] KRISTINA: Tell me, how should I think about organizations that have a website? Because this happens to me a lot with very specific companies that I know I'm picking on them. I'm not going to get to name their names. Cause I think it's not a nice thing to do, but I go to their websites, and I like to visit the Croatian websites because I can easily see what's been translated versus not. And what I find fascinating is when companies translate some of the content on a website but no other content. And it's almost confusing to me. As a user, I can read English and Croatian, so it doesn't matter much, but it seems like it would be; it's just bizarre. Like it's not hard, but it's a little bit jarring to me. It's just that it's not a seamless experience from a pure marketing perspective. Is there an advantage to translating and localizing only some of the content and others? Should we be translating our offerings, but maybe not the news? Is that a good practice, or how does that work from your perspective?

[00:20:00] HILA: Alright, it's a very, very interesting question. I think it's really a matter of budget and demand. Your case is different because you're both English and Croatian. And so you can tell the difference, but think about the Chinese client who comes to your website. And he sees the homepage in Chinese, what solutions do you provide and about the company and for him, this is the website. He does not know that if you mix English, you would see and you want blog posts and case studies. So that I think that assumption of the marketing department or the CEO is that if you do not convert a lead from China, it's enough to communicate with them and to make sure they know what we do before they contact us. And the content we'll have on the English websites is amazing for SEO and organic traffic and for, you know, customer conversion and engagement and education. If you ask me, they do not localize the entire website, and you do budget concerns other than I don't see any other reason not to localize everything. Sometimes it's enough to have basic pages. And I like to have their part of the company, like people, they get to know what you do without the tons of blog posts that you publish every week; it becomes really costly when you localize every piece of content; it goes on and, on the website, I don't think it means they disrespect their audience. So, it really depends.

[00:21:42] KRISTINA: And this is the best. It does because I think you're on the heels of precisely what I was thinking. Yeah, it makes sense to me not to translate all of the content, as you said. And I've ever worked with clients who say things like, well, you know, we don't have a large presence in the Maldives, let's say, right. And so we're not going to translate content at all. And I always respect that decision because I think it's smart to have marketers who understand what their priorities are. Otherwise, if you're trying to be everything to everyone, you're nothing to anyone. So, you have to be smart and go back to our strategy. And so, the translation localization answer that you gave made sense. The one thing I'm wondering is, just from your experience, is it a good practice, or does it matter? Is it situational to have a company may be that translates and localizes 70% of the content and leaves it perhaps in Spanish, and then they still have links to blog posts that are in English? Like, is that an okay thing to do? No, please tell me, tell me, this is what I want to know.

[00:22:41] HILA: I think this is not the best practice because when you are on the Spanish website, you do not want to read English. And if you mix these two languages together, that's not the best part. It's better not to localize, because then really think about this user that you did pay complaining Spanish, you pay money for your website. You found you through SEO, but then if you're providing a confusing experience or if he wants to read something, but he can't. So, in my opinion, I think that's what kind of curiosity. I was like, what's your ticket. But I think for us, I think 70% is okay you meet your, you know, your goals and your audience and the level of demand that you're getting and the end of the day, it's a question of, you know, investment and returns.

[00:23:32] KRISTINA: So I said, I wouldn't ask you this, but I'm going to ask you anyway; you can decline to answer, but I'm curious when you think about globalization and localization, so both who are some of the best folks to look at, who's the gold standard, where you are saying, look, I really respect this company or that group or that, that institution, because they're doing a good job. It's somebody doing a really great job that we should be watching and striving for, or is it the case that we're all just at different maturity levels, and nobody really has this in hand yet?

[00:24:09] HILA: No, I think many companies take these things seriously, and they have really large things depending on the localization of the company. I guess companies like Airbnb and Spotify and Netflix. From our world, HubSpot is doing a great job with a large team of localization managers. And also, there are many Israeli companies that are investing in developing a good localized presence. We are working with Lightricks is an Israeli company. They have this app called Facetune, and they are actively present in many, many markets, and they're localizing everything from their website to their apps to their YouTube ads. And I think this is how it should be done. And there are many brands that are doing a good job. And I think the level of attention that it gets is growing every year.

[00:25:16] KRISTINA: Do you see that there always has to be a large in-house team, or is it possible to do a good job with finding the right partner?

[00:25:26] HILA: Yeah, it really depends on how much you outsource, how much you trust your partner. And there's also lots of work to do internally, you know, even the decision wants to localize, how to do it, creating the content in the first place before you localize it. So sometimes it's seated in marketing. Sometimes it's growth, sometimes it's operations, and sometimes you have internal language experts that, you know, are there to verify and make sure you meet all of its requirements. I think it depends on the level, I heard of teams of 10, to 15 people internally, even among some of our clients, and sometimes it's a one-man show with a good group of vendors and freelancers around there.

[00:26:14] KRISTINA: So no, no one size fits all solution, obviously.

[00:26:18] HILA: No, because it's such a complex thing, you know, localization is everything really, from the customer service from the IVR, when you call the company to the internal communications. Some of our clients are learning and development managers that are in charge of all the tutorials and training for employees and guidebooks. Think about global companies that have to onboard employees in different places. So are a lot of things, a lot of types of content and materials and needs; it's not easy at all. But then we are here to make it easier for them.

[00:27:03] KRISTINA: And thanks for being here because I think that you're making it easier, not just for them, but also for the listeners that have joined us today, and for me, it is a wild ride. I think it's such a great time, too, think, be in this industry and be paying attention to the opportunities that are coming from not necessarily switching from human translation to AI translation and localization, but really to understand where we are on this journey. And what's the right mix for your brand. It sounds like in your markets and your marketing strategy. So, thank you so much for coming on to explain the world of AI localization to us. This has been great. Appreciate you joining us.

[00:27:42] OUTRO: Thank you for joining the Power of Digital Policy; to sign up for our newsletter, get access to policy checklists, detailed information on policies, and other helpful resources, head over to the power of digital policy.com. Please leave a review on iTunes to help your digital colleagues find out about the podcast if you get a moment.