Peggy Anne Salz

Don’t fire your graphics designers just yet. But you can use AI to automate many of the repetitive tasks that currently drown your graphics designers in drudge work and extinguish their creative spark. Our host and Chief Content Officer Peggy Anne Salz catches up with Lior Barak, Data & Automation Consultant with Tale About Data, to discuss how app marketers can start NOW to analyze and automate creatives in order to speed up production and grow profits. Lior draws from his long and impressive track record building and programming AI systems to discuss the catalyst for his Creative Element Explorer tool, a tool designed to understand and scale the component parts of winning creatives. He also debunks popular myths around the best ways to ramp up AI capabilities in your company, starting with good reasons why you should dial down your reliance on Excel spreadsheets.Peggy: Welcome to "Mobile Growth," the podcast series where frontline growth marketing experts share their insights and experiences so you can become a better mobile marketer. That's what it's all about here. That's why we do it. I'm your host, Peggy Anne Salz from MobileGroove. And on my watch, this series will introduce you to the people who know how to drive growth, either because they are themselves growth marketers or because they're enabling that. And if you're listening often, which I hope you are, you're hearing a lot from both.


But I'm really excited about our guest today because, for me, he's one of the very first...and you'll hear this in the show, one of the very first to understand how we're going to...or why and how we're going to take a more data-informed approach to creatives. But that's just part of the show. I won't give it all away right now. You just have to stay there and hang in there. First, we're going to welcome Lior Barak. He is a data and automation consultant at Tale About Data or data, Tale About Data. Lior, just great to have you here on "Mobile Growth."


Lior: Thank you very much. It's an honor to be here. After several tries, I think it's finally happening.


Peggy: Well, the timing is really good. I mean, I hinted at that at the beginning. You know, we've been talking, you know, full disclosure here, what, three, four years and you always understood what it was about when it was about talking about the creators. We'll get to that in a moment. But you've been like a data-driven marketer from day one.


Lior: Yes. I wanted to say I'm...even a marketer would say more that I'm a data person who understands how to do marketing in a better way.


Peggy: Well, see, that's exactly what you're going to need in 2019 because everything I'm reading is about machine learning, automation, AI, figure out what you need to fix, figure out what you need to automate and do it. So you're in a good place here. I mean, I knew you back in the days when you were, you know, a data person at Zalando, but you've moved on. Tell me, first of all, about Tale About Data or data.


Lior: Tale About Data basically comes to enable companies and organizations in solving one of the main key issues, I think, for the marketing department. We have a lot of data. I'll be getting a lot of it, and we're collecting it from different sources. We're working with a lot of partners. But it's really hard to make decisions for people on, "Okay, where should I invest my next Euro?" Or "What should I do, basically, with the data that I received because, okay, how do I actually understand what is the quality of the users that come in?" And I think that 2019 is also a year that we're starting to move away from the CPI, CPA and more trying to understand what is the profit because we don't do anything for free, at the end of the day, we want to have some profit behind it.


And this is where I'm coming into the picture, and helping people to grab all their data, how we get it in the right way, building the data, understanding how to use the data, because I think that most of the companies, they don't know how to do it. And then I'm taking it from there basically to build it into more machine learning, which had been done to get a better understanding of what they do and how to do better. And then, of course, AI to automate a lot of these processes to build tools that can upload biddings or to build tools that basically allow them to analyze images as we're gonna speak later in the podcast about.


Peggy: I mean, you talk about building tools. I mean, I was at your workshop, also at Mobile Growth Summit in Berlin as well. And, you know, it was all about looking at what needs to be done and then doing something about it. I mean, you're hands on. Do you actually build and program what needs to be done? Or can some of this be taken off the shelf? Or is it a little bit of both?


Lior: It's a little bit of both. So I'm more of the consultant type because I can't code. I don't like to code. I mean...


Peggy: No one likes to code, but a couple of very lonely people do that. No, I'm just kidding, just kidding.


Lior: And then, as I said, I'm not a marketer, but I do love to do marketing. So I'm just in the middle between these two functions, I think, and I'm, on parts, I'm telling them, "Okay, this out of the shelf, something you can take, so APIs to download your data." And then when it comes to aggregate it so I'm more hands-on writing down SQL queries to really figure out that information. And then, from there also, if you need to be going into Python and other stuff.


Peggy: You know, you have this new position, you have this new look, you have a website, maybe not new, but I'm seeing it in prep for our podcast for the first time. What I love about is you lay out the challenges because, you know, I'm sure that some of our listeners, they don't entirely...they know, they're drowning in data and I've seen surveys about that. And we'll be seeing more surveys in 2019 because a friend of mine, John Koetsier over at Singular is also working on research to this effect, you know, just how much are we drowning. So we're drowning, but maybe part of the problem is because we have an over-reliance on the tools that aren't really the tools we should be relying on.


And you call out a couple of these common data challenges on your website. I'd love to hear you rant about a few of them. So I'm gonna ask you to do exactly that for me, Lior. You know, what are the common challenges? You know, what do they need to know are at the core of the problem? My first one, over-reliance on Excel. I love that one. You have it on your website as number one. Why is that?


Lior: Because I think that today what happened is that we're downloading a lot of information and we saving it into Excel files and then we start to do VLOOKUPs with it and trying to match them. And once we arrive at this point, people make mistakes, let's face it, right? And we are not completely machines. And sometimes, it's enough that you put the wrong column in the wrong place, or you just did a VLOOKUP on the wrong way and you'll get completely rubbish results. And if you're gonna take it one step forward, think about that, that tomorrow you're sending for your boss and you report about activity of the website. He did not have the time, you open it... So today is Friday, let's say, he opens it on Tuesday, the data will be still relevant.


I'm not sure. You know, when he gonna open it, it's gonna be already outdated and this is why I think that Excel is a very big sickness that we keep using it, but it's really in the wrong way of doing it. The data is not refreshed quite often. We can make a lot of mistake during this process. And as an Excel expert, I said "Stop using it" because it's really causing a lot of issues. You know how many times I send actually a report out and after I hit the "Submit," I discovered that the VLOOKUP was in the wrong place? And this is exactly why I think Excel is a really big sickness of most of this organization because we got used to it. In university, they teach us to use Excel, but they don't also show us that the data is the most important thing behind it.


Peggy: So you make the point, well taken. Good data visualization tool can save you lots of time, lots of mistakes. Do you have one that you, therefore, you know, build or tweak? Do you have any recommendations? Because we're all on the same page. Listeners are saying, "Yep, I am over-reliant on Excel and I have to stop. That has to be my new year's resolution, perhaps." But what are the alternatives? What do you recommend?


Lior: So my two favorite ones are Power BI from Microsoft, which is really easy to install. It's a freemium, then you can start using it almost immediately. And on the other side, Tableau, which are my old friends, let's call it like that. I'm using Tableau now for almost five years, even more than that. And I think that they're doing a great job. They allow you to connect to many data sources in a really easy way. And as a non-data person, it's really intuitive and really easy to use it. It's a huge advantage, I think, to anybody who wants to use a reporting system.


Peggy: Another one you have is drowning in KPIs. And you as a consultant, I'm imagining that you come to your clients and you say, "Look, this is what you need to focus on." And helping people find that focus. It's one thing to get the infrastructure right, the tools right, the data visualization, but it's another to get the right goals. And that's probably where I would imagine your experience really comes in to play. Could you give some tips on how not to drown in KPIs or how to identify the right KPIs? Maybe you have some tips to share.


Lior: So here basically, also, with my clients, what I'm doing, usually, it's a very nice game. We laying out all the KPIs and all the measures that we think are relevant for us. And then we start at a very high level of, "Okay, what are the questions? What are the common questions we have? How many sales? How many active users? How many people clicked on our banners?" And so on and so on. We put it all together in a kind of a backlog. And then from there, I have three columns. And in each of them, there is a limit of how many KPIs you can put in. So, for example, if you're going on a must have, you can have only three golden events, let's say, inside it. I wouldn't call it KPIs, I would call it actually golden events, which you're allowed to insert those.


So, for example, if it's amount... For me, the golden event will be for user clicked on my banner and I produced a value of 50 cents. This is for me high-level conversion. You can go on... A user actually placed an order that is over 50 Euros. This is another golden event. Another thing that for me is important is that the user can form a paid acquisition channel, for example. And then this will be your three main golden events. And then we go into what is nice to have. So what are basically measures and KPIs that are nice to have? And there, I'm limiting it usually to 15. And then we creating this funnel of 15 KPIs and events that we want to include.


And then the not needed. And here, it's basically the game's started to become tough. Most clients, somehow we arriving always to around, I would say, 50 measures in average that they want to have, some measures and KPIs. And then they need now, "Okay, from the golden ones that we just set up, are they really necessary or not? Are they nice to have or are they must to have?" And we start playing with these two and then we basically move in between the columns. I will be happy to share afterward also the tool that I'm using for that. You can print it everybody at home on an A-zeroed paper and do it that I think will be something nice to try and do in the new year.


Peggy: So it's a tool that you sort of come up with or that you use to prioritize KPIs golden opportunities?


Lior: For me, it's a tool that I came up with and I built it based on my experience talking to people and building reports for the last five years for different companies. So it's given quite a lot of input of what and how people think. And the fact that it's limiting people to must have and nice to have KPIs and measures, it's actually reducing the amount of data that they consume. Because, okay, you committed now that there are...this is the three main events that you need to have. So this is your first dashboard that you need to see every morning when you're back to the office. And then the nice to have, it's when you want to start doing research, right? So if you see that there is a decrease, for example, in the amount of orders that you're driving and that the nice to have should expand to why does it happen.


And it's really to go and do the thinking about what is relevant, what is not relevant for me, and what is the most important data that I need to know on a daily basis? And I think it's changing the concept so people don't drown in this amount of KPIs that happen in most cases, you know, because the CEO has one thinking about how we should measure success, the CMO have another one, and then the performance guy who managing Facebook or LinkedIn or whatever has a different concept as well because he's looking on a different perspective. And from this tool, you're supposed to align everybody on the same page, this is what usually I'm doing with most of my clients as well.


Peggy: That sounds very exciting. I will take you up on that offer, either to have you back for a show just about that or to deep dive maybe in a blog over at But for right now, we have to go to break. So, listeners, don't go away because when we come back, we're gonna talk about the other golden opportunity for many of you in the new year is getting your creative right. And Lior has lots of experience and knowledge and tips to share about that. So don't go away. We'll be right back.


Announcer: If you haven't already, be sure to check out the Mobile Growth global event series where thousands of the world's top growth marketing experts come together to discover the latest in growth marketing innovation. We run events all around the world in cities like San Francisco, Los Angeles, New York, Berlin, Montreal, and more. And our candid panel discussions feature industry leaders, including Facebook, Google, Uber, EA, and loads of other top tier publishers. So, attend our lineup of intimate workshops and gain the in-demand skills you need to stay on top of your growth marketing game. Expand your professional network and build meaningful relationships in a fun, friendly, publisher-first environment.


When and where can you get in on the action? Well, just visit for a complete list of upcoming shows. And if you should decide to join us, and I hope that you will, then be sure to use our special promo code MGSPODCAST30 for an additional 30% off your ticket order. Once again, that's MGSPODCAST30, all caps, no spaces for 30% off of your order. We hope to see you there. I certainly hope to see you there. And please, enjoy the rest of our episode.


Peggy: And we're back at "Mobile Growth," and our guest today, Lior Barak, is from Tale About Data, data and automation consultant at that company. And we've been having a great time, Lior, talking about, you know, what you do with your clients, how you get them to focus, some of the tools and tips which I think is really valuable. But the other opportunity, the other challenge is creatives. Now, you've been looking at this since day one because I recall when we met, you were the first to say, "You know, we've got to sort of break down the creatives into their elements. And if we can do that, we can take a more data-informed approach to creatives." And here we are, 2019. That's exactly what everyone is talking about. So I guess for one, congratulations for seeing ahead. Congratulations for that.


Lior: Thank you.


Peggy: What have you been doing? How have you been…? Because you were approaching that way back. So you probably have some progress to report in how we can actually do this.


Lior: Yeah, so it started basically, I think, something like four years ago when I was sitting together with the graphic designer at my company at LOVOO when I worked back then. And his main complaint was that there is no feedback loop coming from the channel managers. And if it arrives, this is not the feedback that actually helps him to be constructive and actually improves the way that he creating his creatives. Back then, he arrived at the situation he was producing around 500 to 1,000 creatives a day, and it's a lot and it was running on 5 or 6 different channels. And we said, "Okay, so there is a challenge here, there is a problem here that we are using a lot of creatives, but there is no feedback loop. Then how can he know what he needs actually to improve in the creatives to drive better users or to get better targeted?"


And we started to break it down into elements. So we said, "Okay, for example, we're using between five to six different colors in our advertisement." And we started in a very simple way to edit into the name tags of the creatives. And we said each color are going to have the first two letters and we're gonna know this is purple, this is blue, this is green, and so on, and so on. And this was basically how the element explorer was born. This is how we started to analyze the small elements inside the creatives. And then we continued. We said, "Okay, in some cases, we're using, for example, a model, in some we're using a product." And we said, "Okay, so let's edit those as well into the naming of the creatives." And we just added initials that say, "Okay, this is a model, this is a product."


And then we continued and we said, "Yeah, but sometimes we're using..." And LOVOO is a dating app, it's important to say. So sometimes, we were using, for example, blonde hair people or brunettes and we wanted to know also what is the difference in conversions? So people when they clicked on it, did they click more on blonde hair or more on brunette hair? And this is where we started to add more and more elements into our creatives and created a metadata for the creative itself. So whenever this one was produced and was published into one of our partners, we could basically go back and say, "Okay, creatives that using brunette hair models working much better than a blonde hair models." That's for as an example.


And then we started to play with it. And we said, "Okay, what else can we do?" And then we started to add, for example, information regarding what product did we use? If we used, for example, a cell phone or we used a computer, this was also a very interesting thing to do. One of the funniest, I think, advertisement that we had was when we used a baby. And it had a very, very bad conversion when it came to male users. But when it arrived to a female user, it was very high converting. And...


Peggy: That's a great story because it's telling you also that you can also throw in a wild card just to check, which is, I'm sure, what you can do with your creative element explorer as well. You don't just check what works, you see what doesn't work. And you learn a lot in that way as well.


Lior: Exactly. And then this was like the first stage of...I think the analytics for this, how it started. And then in the early stage, we understood, "Okay, we cannot any more trust only clicks and impressions because this is a very low level, right? We want to know actually did it drive an e-store? Did it drive a reattribution of the user? So did the user got to engage with us thanks to this advertisement?" And we then started to combine it also with the metadata that comes from our departments from Facebook and Google, and so on, and so on. And we saw who we targeted, what ages did we target, who were the users? Was it male, females, unisex? And so on, and so on.


And this is created the V2 already of this tool. Then we understood that also somehow we need to create some competition. We need to try to do an A/B testing between the creatives. And this is the third version that we actually started to A/B test creatives and A/B test elements inside the creatives. So blonde hair against brunette hair, which one getting the best...back then, we used chi value for that or chi, I think it's called in some countries, it really depends where. And it was a statistical model that basically just compared A, B, and C, and told you which one of them works the best for you.


Peggy: So that's what you were building. That's what you were, you know, ideating at that point. Is this part of also what you're doing at Tale About Data? You've taken this to the next level? I mean, is this part of your baby or what?


Lior: This is part of my baby. It's not even improved today even more because the skills and the capabilities, I think, of, A, the machine learning tools that evolved since then. And, B, also the artificial intelligence tools that evolved since then allow us to do quite a lot of stuff. So if we're looking at creatives, today you don't necessarily need the graphic designer that producing creatives on a daily basis in large amount. It's enough to have a tool-based Python, based on GO or Java or even a simple JavaScript can do that. And you can produce, I won't say millions, but hundreds of thousands of creatives a day.


There is no more limitation as it used to be in the past. And this is where I connected...basically, I started to think about, "Okay, how do we do it?" And we started to create banks of images. And it's kind of...many people will think about it as a product feed. But basically, if you have an e-commerce, many people have a product feed of all their product, it's very similar to that just for images. I will be getting all the images in one place and the machine based on the information that you have gonna start picking up the different images and combine them with colors. And it can contribute hundreds of thousands of creatives in a very, very fast way.


Peggy: That is very cool. I mean, some people listening might say, "Oh-oh, does it replace the graphic designers? Or does it sort of complement them or free them up to maybe do something more interesting?" You know, maybe it lets them think about other campaigns. I mean, what would you say is the interplay between the automated system and the work of the creative designer?


Lior: So here the game is changing because what happened is, as I see, the automatic machine will be kind of a control, right? Because it's learning from the knowledge it has, let's say, not she, she's not human yet. But...


Peggy: Not yet. No.


Lior: But the idea of it is that the machine will be the control. It's gonna produce the mainstream, things that are proven by the algorithm that are working. And then the graphic designer is actually the challenger. So he should come on a daily basis or a weekly basis and say, "Hey, I know that pink color working now really good, but I wanna change it and I wanna say that purple is working much better." And this is the place of the graphic designer now to come and actually challenge the machine. And basically, the machine gonna learn and say, "Okay, from now on, I see purple is working actually better than pink," and add it into the algorithm. So the machine learning gonna feed it into the automatic machine that is running there.


And this is basically the new role of the graphic designer because I think that we don't need anymore... It shouldn't be like a huge effort from the graphic designer to produce hundreds of creatives. It's actually a wasting of his time. Because graphic designers are very creative, very thinking on concepts people and this is what they should do. They should not waste their time on reproducing the same creatives hundreds of times just because we need it in different sizes, in different location, in different languages. This, the machine can take care of and then the creative part of challenging it should come from the graphic designer.


Peggy: That's really interesting and that's also very positive if you think about, Lior, because it's not that anyone's going to be replaced, they're going to be enhanced or augmented. It reminds me of an event I went to around AI, oh, two, two and a half years ago now, where people were talking about how they work currently with data and how much dredge work it is. You know, it's like, even for architects. Like 70% of the time that they spend is figuring out, you know, will the bridge hold?


It's not like design something really amazing, it's just like, you know, the basics, doing all the dredge work. And here it's very similar. It's the idea that you can reproduce and change and move the creatives around, but you can be, you know, path-breaking and cutting edge because you have time to do it. You can challenge the machine to do it. So actually, ultimately, I would imagine what the results are with your clients, is what, that they are able to push the envelope just a little bit more? Can you give some examples?


Lior: So some of the clients actually improve the amount of creatives they're testing. One of the biggest challenges at the end of the day is to wait for the creative designer to finish creating something for you, then you need to upload it, you need to evaluate it. And with this automation, what you need to do is basically say, "Okay, this is the format of how the outlay of the creative should look like." Then that machine gonna start taking it automatically and it's saving for him also a lot of time.


Because the only thing he needed to come up with is a format in saying, "Okay, on the left side, there should be a 16 or 9 video. And then to the right of it should be another full one-on-one images that are gonna be randomly change." And the machine can take it and start producing it. For him, it's just to design it either in a very simple HTML or in a very simple JavaScript. And from there, they can just take it and start building.


Peggy: So we're at the start of the new year. You're with your new or what appears to me to be new. When actually did you found it by the way?


Lior: Tale About Data?


Peggy: Yeah.


Lior: We founded last year. And it's actually was still when I was working for Zalando, for me, it was more platform of arriving to conferences and speak in an easy way. But then I had a lot of concepts that, like I said, "Okay, what are we doing from here? Because it needs to be shared outside, I think." Because the element explorer, for example, for me, it's a tool that I'm running with four years already and there is a platform that I needed to share this information with people because I think it's important to give it to people as a tool.


Peggy: But I love the fact that you're sharing it here. I'm also wondering, you know, what kinds of clients would be a best fit with you? I mean, are there certain types of apps that you would work better with, or types of clients you would work better with? Just wondering how that would work.


Lior: It really depends on the situation of the clients and what they need. So for example, element explorer, I think, it's something that is more global. It's for e-commerce, gaming, subscription apps and so on and so on. On data side, I think that some of my services are more focused on e-commerce. It really depend on the end goal of the users, but then they can always contact me and we can do an evaluation. So I don't like to waste time with people. I'm very pragmatic in the way I'm looking at stuff. And my viewing is very simple at the end of the day. How can we actually work together and make something good for the organization with the organization.


Peggy: Perfect. So, Lior, how can people best get in touch with you?


Lior: On LinkedIn it's the super easy way, of course through Twitter, it's @liorb. And email as well.


Peggy: Okay. And we have all of that, of course in the show notes and where you also find this episode to listen in to. So, yes, friends, thanks for listening in to this episode of the "Mobile Growth Podcast." A quick reminder to visit for a complete list of our upcoming events. And don't forget to use the very special promo code MGSPODCAST30 for 30% off your order.


We hope to see you there. I hope to see you there. And we also encourage you in the interim, in the meantime, to check out this and earlier episodes of our series by going to It's also available on SoundCloud and coming to many more channels providing you many more ways to listen in. Watch for that, we'll watch for you, and we'll see you soon.