#8 Milica Markovic | How a single Google Sheet can fuel your whole CRM automation

INSIDE CRM with Milica Markovic: Mastering Marketing Automation Through Simplification

Step into the world of innovative CRM solutions as Milica Markovic, a CRM expert at kfzteile24, shares her approach to managing customer engagement across a 5-million-product automotive parts marketplace. Drawing from her experience at Smava and her current role, Milica demonstrates how to transform complex marketing automation into streamlined, efficient processes.

Her team shifted from 80% routine tasks to focused innovation through strategic automation. By treating messaging as code rather than design, they implemented a system using Google Sheets and CSV files to manage complex personalization at scale. Through smart segmentation and strategic use of Bloomreach, they achieved one-to-one personalization without overcomplicated solutions.

The episode reveals practical insights into building scalable automation systems through asset farms, master templates, and recommendation engines. Milica's team achieved 1.5 billion template combinations while maintaining operational simplicity. She emphasizes giving teams freedom to experiment, leading to successful initiatives in contextual personalization and AI-driven content creation, while maintaining focus on revenue growth and customer lifetime value.

00:00 Introduction to Milica and Her Background
03:15 Simplifying Complex Marketing Automation
07:45 Building the Google Sheets System
12:30 Achieving One-to-One Personalization
16:20 Team Innovation and Experimentation
20:15 The Future of CRM and Marketing Automation

Transcript

Speaker 1 0:01

Welcome to Inside CRM, the podcast where we explore together the tools, trends and tactics to improve our customer relationship management. I'm Jessica Janssen, the host, but also CRM expert myself, with over a decade of experience. In each episode, CRM professionals join us to share their knowledge to sharpen our CRM skills. Let's get started.

Speaker 2 0:25

Yeah, we're doing it now. Here is my hero to the rescue. So background normally someone else would have been here, but he got sick. So you came in at six and I was like I have this problem. I'm missing one person, one content, and she is just jumping in casually. We will do a chat and maybe start with your intro. Awesome.

Speaker 3 0:52

So, yeah, I'm Milica, I'm from Belgrade, but I live in Berlin since seven years and I've been working in Smava, which is a long comparison portal, doing all the omni-channel stuff that goes into it implementation of the tools, et cetera acting product owner at the time, and now I'm in KFC, tyler 24. We sell car parts, so not the most fun product ever, but from a CRM perspective, it's a very interesting problem to have on your hands, because you have 5 million products and now you figure out personalization and how that's going to look like. So what we've done and that's what Jessica and I were discussing we can talk a little bit about marketing automation and how we got there.

Speaker 2 1:38

Even though she has no background, I'm going to explain it right, but it's also a crazy journey because she explained it to me already once. It was really impressive. It also felt a bit like a really structured and complex setup that you have.

Speaker 3 1:53

Yeah, I don't think it's complex. I think I read somewhere that one of the greatest assets that's going to be for us all in management, the most valuable asset we would have, is to simplify the complex issues right. So you can build something incredibly difficult, but there is no use. And this is a good intro because I wrote down the note. So I have a team and we have a rule and when we started working on marketing automation, I told them currently you're doing 80% of boring stuff. Currently you're doing 80% of boring stuff. It's a nice introduction, but it's needed and you have 20% to do cool stuff. If you do cool stuff in a right way, as in release as much marketing automation as possible, you will have more time for cool stuff and you will reduce the boring stuff. So that's one principle. And the other principle is see message as a code, not as a design. Imagine it like SQL or it is HTML and CSS, but imagine it like a set of codes and it's a text and in the text you can embed stuff and you can take out stuff.

Speaker 3 3:03

When we change from email, for example, because email is the most complex one, when we change it from, this is banner and then there is copy and then there is a button and then there is this color and that color, but what's in the background? We actually managed to create like a full automation and we're talking I don't know 30 segmentations, different channels email, smss, push notification, in-app message, etc. So we've done three things and you can, if you've done that, challenge me please on this one. So we've built like a google doc. Take this, no, I'm good waving. So we took, I wrote it here, but maybe someday I will prepare slides and you can go into details, right, but we took one.

Speaker 3 3:46

One part of the template is everything that comes from brand and the content that comes in. So we are talking banner, we are talking copy, everything that is done on weekly level. For that part, we provided a google doc for brand with the exact structure that we need and every field in that template is connected in that template for brand, is connected directly and mapped into a master template that we have in our tool. And then the second pillar of it is all the other elements you have within the templates, right? So we are talking USPs, we are talking delivery, we're talking all these other extra stuff. So we built an asset farm and we have hundreds of elements that are directly mapped into the design elements of the master template, so we can turn them on and off by selecting them in that Google Doc that we provided the automatic one.

Speaker 2 4:44

And how do you connect the Google Sheet with the tool we?

Speaker 3 4:48

have the Google Sheet that is being exported into CSV and then uploaded from CSV.

Speaker 3 4:55

And that's it. You don't touch a thing. So the third thing is the recommendation. So you turn on an element. That element, in background in the tool, has already all the fields connected to the recommendation center, which is upsell, cross sell, last seen, seasonal stuff and whatever you want to do. So basically, this google doc will decide element one, two, five, seven, fifteen, uploaded connected to the segments, because you're able to in google doc to basically say these are the segments that I have.

Speaker 3 5:27

This is the copy, this is the banner, this is the voucher code. You have standardized tracking. You download the CSV, you upload it into not even a joke, you upload it into a tool and then you spend a couple of hours just testing if everything works. Maybe a couple of hours because you want to get the feedback from customer care and stuff, because selling car parts is not easy for us non-car part people, and that's it. And if you have a tracking, then you have in GA and then in Tableau on a segment level, all the clicks, all the opens, everything, all the conversions, and now we just do cool stuff. How many google sheets do you have?

Speaker 2 6:07

we have one this is all, one and several tips.

Speaker 3 6:11

I'm lying right. It's two, yeah, one, one google sheet. One google sheet is the one that we provide for brand and pricing and they're filling in the, the brand, and the second one is the asset farm. An asset farm I don't know if you have that concept. It's about centralized. If we ever do this, I can put on the slides, right, but it's like a centralized database of all the it's called. It's. Basically it's a google doc and it has this is the headline, this is the copy, this is what's in the cta and this is an image that goes with it.

Speaker 2 6:45

Those like are filling in the elements but also the tracking, the segment, so you're also controlling it via that right yes, we have these master templates.

Speaker 3 6:56

It took us years to get there, guys. It's not like we came out brilliantly, it's like trial and error. I do crM for many years. So, yeah, you have two. Actually, the one that is an asset farm, you just add when you have something new, like you change something about the delivery, you add that information, edit that information, but that remains. Once you build it, it remains stable, and that's as I said.

Speaker 3 7:22

It can be about some categories. It can be about some products that you want to push. It can be a catalog of products that you can push as a recommendation, because everyone thinks recommendation engine has to be the smartest thing ever made. But you can fake it. How do you fake it? How do you fake it? Easily, you go to pricing and you tell them give me the list of top 100 products. You have a discount of 20%, because it's stupid to show one euro discount, right, yeah, so they send us that. And then we pick oh, this one works, this one doesn't work.

Speaker 3 7:56

We put in that same asset farm doc. Right, we have automation for that. We put into the same Google doc. We put those products. It automatically populates events like elements. It just ships out, and you can do it also on a segment level, so you can say this is the product portfolio for segment one.

Speaker 3 8:15

Segment two, it's essentially think about CRM. If I'm like, it's very confusing because this is not a structured presentation, it's an improvisation, but if you think about every message that you send out, it works for push notification, it works for SMS, it works for everything. We use chat GPT to push in the content because we were bored, we wanted to try it. I would say, if you think about it as a piece of code, that's the first important thing and you can take in and out anything from it. And the other thing is you find the way to centralize as much and reuse as much. You're going to end up in a very automated flow that is easy to maintain. It depends on how you write your copy, though I don't know. Do you guys all write copy internally or do you have copywriters like in other teams, like in brand or something? You have it in brand, yeah. So imagine that you become a product team. Actually, we lead CRM as a product team, not as a marketing team.

Speaker 2 9:20

But why was it necessary to set up the Google Sheets Normally? You?

Speaker 3 9:24

would like to. That's the way the tool works. The tool is Bloomreach, but it's so interesting how that works and it has to be the most impressive sheet I've ever seen. It works on CSVs, it's awesome. So we faked it till we made it in Marketo as well, so we played with Marketo. Made it in Marketo as well, so we played with Marketo.

Speaker 3 9:42

We previously used Marketo in Smava, so there were times when we were testing the upsell and we didn't have recommendations. So we created the business logic. We went to BI. They put it into Tableau, so Tableau was running the recommendations. Not even a joke. We would download that and put it before Marketo into what was it Optimo? Then it became EpiServer, so EpiServer was run only on the manual list. We did it with EpiServer. If you imagine, everything is Excel, like it's code and it's a VLOOKUP Excel. Easy, just apply the principle and you will eventually get there, because recommendation engine is essentially any machine learning like. It's a very buzz of a word, right, but any machine learning is very capable VLOOKUP, like incredibly capable Excel. In Marketo they called me the feature abuser. If you understand how something works and what it does, you can reuse it for whatever your use case actually is.

Speaker 2 10:50

Do you have scaling problems with it? What is the performance of the setup?

Speaker 3 10:54

No, how many segments you want to have, right? So we are achieving one-on-one personalization. Back in Smava, we had 1.5 billion combinations in one template that would go out. Why? Because we had recommendation logic. That was easy. Because you can have loans from 1,000 to 120,000. So your upsell and cross-sell are definite right. And here we have full personalization.

Speaker 3 11:19

Because the way that people think about one-on-one personalization is that you must have an engine that is going to run somewhere and recommend something to you, but in fact, a lot of customers in terms of some user characteristics. Essentially what my approach was like. What if we had five different segmentations that each have 10 different segments inside? And then what are the parts that I personalize, which can be done one by one? So in car parts, if I'm driving Toyota and the last thing I searched was this particular product, displaying that category or a product within a template is already one-on-one personalization. And if you have based on these five different segmentation, if every element is adjusted to your own segment, that means that you are going to have that many options, right? So it's five segmentations and 10 segments. So it's five on 10 on 10 on 10 on 10. It's insane. Yeah, like it. It can work quite cool. So don't think about the simplest way to push a decision to push that information through, and then you're easy.

Speaker 2 12:53

Have you now increased the time that you spend on the course stuff? Because you were saying in the beginning 80-20?

Speaker 3 12:59

20, yeah, my team has a four-hour meeting on Fridays without me and they're talking about cool stuff. And then they come up with all these ideas. And then they come back to me on Monday or Thursday because we have, as I said, we function as a product team, so we have the stand-ups and we have the standardized ceremonies. They come to me and they're like we've discussed, we want to try this out, or even it happens oh, we are in a show fix. Oh, I forgot to tell you. I run a contextual personalization test last week and I'm like, okay, what was it about? And then they show you these things because they feel very much.

Speaker 3 13:40

I made a mistake in my previous company, right, and I was not a good lead. I had to learn how to be a good lead and how to motivate people to actually love what they do, and what I learned with my team is that they like to play. We all have children in us, so they have full freedom to try out things, as long as they are not very negative on the revenue. So I even am like you can try it out. It's not going to work and we're going to lose some money, but if you can find a way for us to win that money back. That's a win Because you remember your first campaign where you failed. I remember mine. I lost money of course we all did exactly.

Speaker 3 14:18

You lose money. It's not like I lost revenue, it was gave away money. I was like I was not bad, I was worse than that and I cried. I was so scared. I lost, like in Serbia, money. We're talking three thousand,000 euros. So it's funny. But I remember how scared I was and how horrible it felt. But then I was making millions for the same company like a couple of months later. So it's always a gamble If you don't test and if you don't try and if you don't fail and everyone's saying fail fast, you don't have to be stupid, but trying out things makes sense.

Speaker 2 14:58

So what fun stuff are they working on currently?

Speaker 3 15:01

So there is this contextual personalization that they've released behind my back, which I support. I was not pissed and that's very cool, because the whole abandoned journey abandoned cart journey for us looks very different than it does for the rest of the companies. Usually it's one email that shoots out. We're like we're going to make this a journey and it's going to be omnichannel and it's going to be everywhere and it's going to be awesome. So we have, in each one of these nodes Now the team implemented three different templates, and so the engine based on customers behavior and clustering of the customers. So there is Bloomreach is a smart tool as well, so it's able to choose hey, BRC, and we don't control it.

Speaker 3 15:39

And you don't have a winner, because a winner is every one of these templates. It works for some group. So basically, the third thing is there is no such thing as control in personalization. You have to give up control and that's fine. Mistakes will happen and that's how you will learn. Yeah, they do that. They do personalization. They did the chat GPT versus editors desk, which editors didn't like because chat GPT won every single time.

Speaker 3 16:12

We don't know how. So we upload a little bit of data inside like what's the open rate, click rate, stuff like that. Then you will get based on that performance and behavior. You will get some recommendations. Then we check if that makes sense and then my team just went and tested it, implemented it. It's happening. Additional customer journeys. They're like oh, I was thinking I would like to send out an email on this node in customer journey. It just makes sense. And you spoke about that, which is very cool. We have the whole journey mapped out and then my team looks at that journey on these four hours, without me I would love to attend this meeting.

Speaker 2 16:50

What are they talking about? I always have a question, maybe about you.

Speaker 3 16:53

Yeah, they're like oh, the's the word, they're super sweet. They're super sweet, but yeah, so they map out the customer. We have mapped out the customer journey. We already put in some ideas. They go back to the customer journey, because this is where the truth is, and then they're like maybe we can try out sending here an email like after the second purchase. Like after Try out sending here an email like after the second purchase, like after the first purchase, do we push them immediately to buy a second product? Why not? And then we beg them buy another one. It's going to be awesome.

Speaker 3 17:26

And you do all of these things right. The way that you get consent, like I have some tips for you, if you want to. The way that you get consent sometimes you need to be very sneaky about these things, right, not illegal, but sneaky. And they also deal with those topics. They're like how do we get the customers who are running out of consent by the legal ways to reconfirm their consent? So then you will send an email hey, you're about to get out of consent, but if you want this discount and this, then you are going to get and confirm the consent. We're going to give you this massive discount for the next purchase. And then they're like yeah, sure, give me emails, but at the end customers we have 70% returning customers on the company level, 70% returning customers on the company level and we see that a customer who has a consent is more than two times more valuable in CLV than the customer without consent.

Speaker 3 18:23

There is no discussion Companies are underestimating CRM. There is no paid marketing without CRM. There is no commercial without CRM. There is no scaling and growth without CRM being profitable. Yeah, this is what we do. We just play around, release stuff, that's it.

Speaker 2 18:44

You mentioned actually the topic. Crm is underestimated and I sometimes also hear this discussion that you have a global control group to prove the impact of CRM. What's your experience there? Did you have to prove it in such a way? How did you set it up? If you did that?

Speaker 3 19:03

I had to prove it so many times. It's even sad. Yeah, we were just talking about it. You're paying for a tool for CRM for the whole year, the half of a monthly budget for performance Usually that's the count in the company and yet you have to prove that CRM works right and that segmentations work, and all of that.

Speaker 3 19:24

And we don't have a global test group because we compare to the non-consent people. So it's very clear when you compare these two groups, every test that we run against of course the control group. So you have to test it and test it again and you have to prove it and prove it all over again. And that sucks, yeah, to be honest. Yeah, sometimes you feel very misunderstood in companies. But, to be honest, our time is coming. It's true, because the acquisition costs are so high. Google is now an automation. Essentially, the customer retaining a customer, maintaining a good relationship with customer, will always be a king. So I see chief customer officers in next couple of years popping up left and right and doing these things. So just hold on, don't change your career. Should we stop it?

Closing Remarks and Call to Action

Speaker 2 20:16

with this positive message. I feel like it's a perfect ending for it. Yeah, yeah, yeah, it was so great to have you and that you took the spot and improvised totally. Thank you so much. Thank you, guys.

Speaker 1 20:31

You made it this far. I truly appreciate your time and attention. If you enjoyed it, please leave a review on Apple or Spotify, because it helps me to connect with more CRM professionals like you, and don't forget to also share the episode if you think colleagues or friends will benefit from it. Subscribe to the podcast and visit insidecrmio to sign up for my newsletter that comes out every week filled with CIM tips, stories and resources. Thanks for listening in and I hope I see you around next time.