Big data for marketers: Why, how, and where to start

In this episode from our SuperSummit virtual event, Khrystyna Grynko, Cloud Customer Engineer at Google, takes you through all the steps to start utilizing big data as a marketer. She explores how to define your use cases and choose the right tools to unlock the power of your data within your marketing team.

You'll learn

  • The steps of utilizing big data as a marketerm
  • The differences between data lakes, warehouses, and lake houses
  • How to pick the right tools

Subscribe to the Marketing Intelligence Show

Learn from Supermetrics' experts how to use data to fuel growth and maximize the ROI of your marketing spend.
Subscribe to the Marketing Intelligence Show

Khrystyna Grynko:
Hi, everyone. You're in the talk, “big data for marketers,” but I think it will be interesting for everyone, for developers as well, because sometimes you need to work with marketers to help them implement what they want to implement.

So, I will quickly introduce myself. I'm Khrystyna. I'm working as a cloud customer engineer at Google, specialized in our data analytics services, customer engineer. It's like presales engineer if you don't know what customer engineer is. And I'm in the sales organization and I'm working with the customers to help them to start using different cloud services.

And I used to work in a digital marketing agency and I started first as a digital marketer. And then gradually I started working with data quite a lot. And then I was super interested in the cloud, and I decided to go into the cloud.

So I'm doing the facilitation for the unmarkable sessions. It's a very cool initiative that actually helps people to learn how to self-promote, how to talk about their achievements. And I'm lecturing, mentoring, volunteering a bit, writing a little bit and yeah. And I suggest that we stay in touch on LinkedIn. And after the session, if you have any questions, we will have the Q&A at the end of the session.

So, I wanted first to talk to you about the technical marketer and the need of non-technical marketers. I remember this article written by Simo Ahava. He is quite an expert in our digital area. And he wrote this article about the need of the non-technical marketer, saying that hey, folks, we're actually technical. If you're a marketer and you think you're not technical, it's quite impossible these days because who says marketing today says digital marketing.

Digital marketers use pretty sophisticated tools and know how to track stuff, how to put sometimes JavaScript code on the website to track conversions, or how to add some UTMs to URL and to shorten this URL to put it in the blog post. So we marketers, and I say we, because I used to be a marketer as well, we do a lot of technical stuff, but we still sometimes consider ourselves non-technical, saying, oh, it's too technical for me. I won't try to understand this or that because it might be too technical for me, but it's not actually the case. And you should be reassured because we all think like that.

I used to think this way about technical skills. They became increasingly. important for us marketers. Why so? Because this whole buzz was the Gen AI at the moment, everybody wants to use generative AI. And all of a sudden, there are plenty of experts in this. We kind of realized that, Gen AI, we want to use AI in our data, but we actually want to have some data as well, right?
We need some data to work to implement some Gen AI techniques.

So how about data? So we're getting back in this talk to this term, big data and big data for marketers. And I wanted to explain where you should start to kind of get familiarized with the subject and why it's important for technical marketers and non-technical marketers.

I wanted to kind of show you as well a nice article written by Juliana Jackson, quite an expert and influencer in our field as well, because she made this journey from marketing from growth hacking to the technical part. And she wrote this technical marketing guide to help those who are transitioning as well and want to know more about different IT skills. So I really recommend it to you.

So why should you be technical as marketer and why is big data is important. I shouldn't even explain that, right? We all know why. We know that we need data to do the AI. We know that we need data to personalize experiences for our customers on the website. We know that we need big data to target our audience better, to show the relevant ads, to analyze area of our ads. We know all that, but sometimes we just don't know where to start.

I quickly went to our customer reference website for Google Cloud and about different marketing analytics cases just to show you a few examples of how actually customers use marketing analytics data for their marketing analytics and for their marketing in general. And there are some who use data to kind of improve their customer behavior analysis. There are some who use data to understand better what the customers want. Some who use it to calculate better ROI, because sometimes we have a hard time as marketers to understand what actually brings us money, what actions bring the company money, and how do we prove that everything that we invested in all this advertising around it, how it actually helped us and some use big data to personalize customer experience, but there are much more other cases.

There are so many tools for us as marketers. They all fight for our attentions. They’re like “use our tool, do something, buy our solution.” They all want to help us. They all create amazing products to help us automate stuff, to do it better, to do it in the best possible way. And there are so many of them, and we have a hard time to choose the best solutions for this or that specific problem. It's quite challenging. And sometimes we also say, but I lack technical skills to evaluate if a tool is good or not.

And sometimes we just use something that we know, something that we heard about. Something that, for example, we saw on LinkedIn, somebody posted about it. We maybe ask some questions, and we know, okay let's try this tool, for example. Or we just compare by price and use the cheapest one.

It depends on our budget as marketers, right? But basically, when we talk about data and especially big data, what's important, for example, for me, when I did this transition between non-technical and technical marketer, and when I started using more and more data for my marketing actions, I realized that actually the things that are important is how do you store your data and clean it?

Where is the place to store and clean your big data? What do you do to activate this data? Basically, that's it, right? And what I mean by activating the data is making some magic with it, make something great with it. Use machine learning AI to do some cool stuff or maybe to create lookalike audiences to create some super personalized experience to recommend products, et cetera.

So is there any place where I can put all my data and then connect my super-duper cool marketing tools to this data in order to get some value from it? And sometimes you think oh, there are so many things we need to know. So many technical things. And we imagine some kind of the toolkit of the full-scale developers saying, I probably need to know everything right? Need to know all of that.

But basically what we need to know to do big data for marketing, first thing you need to know is what is data warehouse, data lake, and Data lake house. There is usually quite a confusion between these terms.

Data warehouse is where you store very organized structured or semi structured data, usually tables. It's the place where you store tables with data. Data Lake is the place where you can store everything that’s data, right? Images, PDF documents, CSV files, etc. Plenty of different documents or elements that you have that can be data, you can store them in a data lake.

Data Lake is usually the place where you store stuff that will be probably used in a transformed way. In the data warehouse, for example, you say, but how can you use images, for example, in data warehouse? Well, imagine that you ask some ML service to go through images and tag them automatically with the words saying, these are the images with dogs. And for example, in your data warehouse, you store one column with the link to the images and one column with the tags. So now your unstructured data becomes structured data. And in a data lake, you have your images in data warehouse. You have the links to the images and the tags that are used.

There is another term that is quite often used these days, data lake house. So where you combine both saying, okay, now the tools are so powerful that you can do machine learning and some sophisticated stuff on unstructured and structured data at the same time. So sometimes you can meet this term as well.

So okay, now we understand the terms to store your data, how do you explore this data? Our human eyes have a hard time exploring big data, because somebody once told me that big data is when you have to scroll a lot, right? It's just a huge amount of data. You scroll, you scroll, you try to understand what's going on, but it's very hard to understand what's happening.

And this is why data visualization was created in the first place. We all know that it's quite handy to see a nice graphic that shows the numbers go up or go down. So a good BI or visualization tool. And I made a distinction here because visualization tool just helps you to visualize things and BI, or business intelligence, usually helps you to do some sophisticated analysis to convey the combined data between tables, etcetera.

And there is SQL, the language is not really a programming language, but it's a language that helps you to communicate with the database or data warehouse. And at the basic level, it's pretty easy to use, even for the non-technical marketers. To know some basic SQL can help you when exploring your data in the data warehouse, for example, and some BI or visualization tools use a little bit of SQL as well. So I would say SQL knowledge, at least some basic SQL knowledge, would be quite handy as well.

The third part is to know how to feed your usual marketing tools with the data. For example, prepare your audience in the data warehouse and then export it to your emailing tool or to export. For example, using machine learning, classify your audience into segments and export them, for example, to advertising tools.

So how do you connect your marketing tools? With the data and usually the sophisticated marketing tools on the market, they already have connectors to the most known data warehouses on the market and they just tell you just give us access to the data and we will do the rest.

An important note here. Let's be serious. Don't forget about privacy. Really important. So use the tools that promise to respect the privacy of your customers. And something that I really like is the nice schemas that show what's going on. So imagine on your left, you have a bunch of sources, different sources, your web analytics data, your CRM, social media data etc, that you connect to your data warehouse and you send this data regularly.

There are two small words there, batch and stream. What it means is that, it should be familiar to you, but batch is when you send something once per day, for example, you collect the data in your advertising tool and you send it once a day. In stream mode, it's streaming all the time. You send data every time there is a data point, every time there is an event. You send the data.

It's got it can be quite handy when you need to analyze data in real time. For example, I don't know if there is a sport event and you want to analyze how many people are watching this game at the moment and you don't need to know this data tomorrow, you need to know it now. So maybe you will want to make sure with your team that you're loading this data in the streaming mode. Then you connect your BI or visualization tool to the data warehouse.

Most of the known BI tools or visualization tools on the market connect quite easily to the known data warehouses in the market or you connect different in ML, so machine learning or artificial intelligence platforms to your data in order to do some magic, even more magic that you already did some data warehouses and data lakes and data lake houses. They give you the possibility to do ML or AI inside these services and solutions.

Once your data is ready, you can kind of distribute the reports to everyone and you can use it in the marketing tool as well. So just put it into schema.

I really wanted to show you the process of thinking of the marketers when they want to use big data. Let's break it into pieces. So first of all, always start thinking about the use cases, because sometimes we want to buy a nice marketing tool. And then afterwards, we try to tailor our needs to this tool. We say like, okay, I have this tool. It has its limitations. Let's use it within its limitations. But it's better to think first about what are your use cases and priorities as a marketer or analyst.

Do you want first to use data in order to recommend products? Or to do better ads, or to target better the audience? So try to think of the several use cases first. Then think about the place where you will store your data. Do you need to acquire a data warehouse solution, or maybe you have already one in the house?

And in this case, I would suggest that you, as a marketer or analyst, go and talk to your IT team. Probably there is already a data warehouse used in the enterprise, and maybe you can take some part of it and start using it for your marketing purposes. This is a huge problem of big organization and even sometimes medium organization that the departments don't communicate with each other. So go and ask, maybe you have already a nice data warehouse and data visualization tool.

The third thing is listing your data sources needed for your use cases. Where do you need to take data from? Will it be Google Analytics, Google Ads, Facebook Ads, Bing Ads, etc. What are your sources of data? Maybe some emailing, MailChimp, etc.

Once you listed the sources, define how to get this data, the data from the sources to your data warehouse in order to consolidate it, to put it together.

Usually for the known tools on the market, there are connectors. There are connectors, for example, the Supermetrics connector that we all know. And, they help you to connect your advertising, emailing, et cetera tools to your data warehouse quite easily. You have always the option to create your own connector. You have the resources internally. So think about it, maybe you have some cases where the customer has their own in-house CRM and obviously there is no connector existing on the market, but there are always ways to download, for example, the CSV from the CRM with the data one time per day and send it to your cloud data warehouse. There are always the solutions for that.

Define whether the data is usable as it is, whether once it's in the data warehouse, it's already nice and tidy, but it's rarely the case, of course. But it can be used. Or whether you need to do some cleaning and data preparation.

And if you want to do it in data warehouse or maybe already in a visualization tool, then the next step is to choose your BI visualization tool. But there is another step, define what reports you need. And probably you will argue with me saying, yeah, but maybe we should think about what reports we need, and then we should choose the visualization tool.

And you will be right. So, because sometimes if you need some kind of sophisticated, unusual reports, probably the tool that you will choose for that will be quite limiting, or maybe it will not have the functionality that you need. So you, you might want to decide at first what kind of reports you will need, dashboards and reports, and who will need them.

Maybe there is the management team that they want to have the reports presented in a specific way, and you want to make sure, for example, that this BI tool helps you to download the report into the slides format and or schedule that PDF sent from the report automatically every day to the manager's email address.

So once you define the use cases, choose your BI visualization tool, find how to leverage the console data you have. Now that you have this data, you have this gold stored well prepared inside your data warehouse. Now you can connect your marketing tools and quite usually they are already ready to accept your data to start doing things and to say, hey, if you give me the possibility to connect to the data warehouse, I can help you create the audience, for example. So then how can you leverage the existing tools that you use with all this data that you have inside your data warehouse.

And so for the big data for marketing starter pack, I would say that that would be the understanding of the concepts mentioned in this presentation. Read a lot about the use cases, read about what others do with big data.

There are plenty of great articles, big data for marketing purposes that explain how different companies use big data for their marketing. So inspire yourself. If you cannot come up with the use cases, don't hesitate to go and see what others do.

It's quite obvious as advice, but I like this one – sometimes to understand what tool is nice or not is if it’s something that can respond to your goals. There are a bunch of videos that you can search for. Start typing “introduction to Lucre studio, for example, in order to understand and get some quick tutorial about the solution to see how to use it. It will help you understand the basics.

And I really suggest to learn some basic SQL. I wanted to share some nice Google educational resources with you. The cloud skills boost used to be called quick labs. And it actually gives you a bunch of different labs that you can get step-by-step guidance where you need to click, what you exactly need to do and for what. It's really educating and I like it as well because it creates a kind of sandbox environment and you don't have to use your tools and your data and your environment. It can create something for you in order for you to train in the environment that is created just within this lab’s time limit.

So and on this cloud skill boost, there are some labs that require credits. So subscription and there are a lot of free labs. And I think I selected only the free laps for you here and, of course, because we talk about big data and there is these amazing buzzwords at the moment, generative AI, everybody's talking about it. There are some labs as well on the cloud skill boost website. So a small bonus for you if you want to become even more technical than you are already, and it's actually interesting because it uses ML and AI with the data warehouse solution, to get familiarized with. Warehouse data solutions and AI ML at the same time. It's pretty nice.

I think we can go to the Q&A part.

So when should an organization look into data warehouse? Do you have in your condition in terms of which data warehouse is good? Well, I'm a bit biased here because I work in Google, and I find that BigQuery is super amazing as data warehouse solution. So I would definitely say BigQuery. But of course, it's up to you to evaluate.

There are top performers in the market. But yeah, I am biased. I actually started working at Google because I really loved BigQuery and as a marketer, I used it quite a lot. But the question about when should an organization look into data warehouse? Now. Whenever you have more than two sources of data, I would suggest go ahead and start doing your data warehouse. You will thank yourself later. If you have two advertising platforms, you definitely want to combine this data together to do some nice analysis and some automation. So, start now if you have more than two data sources.

So next question, what are the best tools and resources to start learning SQL and become proficient?

So when you choose the data warehouse that you will work with, I suggest that you search the course. For example, when I started working with BigQuery, I found the specific course that was SQL for BigQuery and I really loved it. It was super basic and it really explained how to do it inside BigQuery because SQL has dialects as in normal language. So choose your tool first, then just Google the course. There are plenty of different courses, like for example, SQL in Snowflake, SQL in BigQuery, SQL in Databricks, et cetera.

Okay, can you talk more about data warehouse, data lake, and data lake house? How can you choose between so many options? So usually you don't have to choose because data lake will be something between your data and the data warehouse. In data lake, you can send everything, you can send PDFs, you can send images, you can send CSV file, etc.

Everything that is unstructured or structured can be there. You can, for example, say that, hey, I have my bucket with all the data there. Once I clean it and structure it, I can put it in a data warehouse. I would say that usually you need to have both or to have, for example, some lake house solution. But yeah so it's usually both.

All right, so another question, in any organization without the data BI tools mindset, it's always difficult to invite them to stop using Excel. How would you recommend educating teams on using BI tools?

Well, always people are scared of change, right? Changing the tools, learning something new, sometimes it's a bit scary and we think it might be complicated, so you need good training to show how the product is amazing and easy to use at the beginning, but also some guidelines and some kind of, hey, now for the report, we only use this. You can do Excel on your side if you want, if it's easier for you, but then put it in the right form in the report to have some specific guideline or on the level at the level of the company saying that for the reporting we use this tool.
But make sure to explain the solution in a friendly way, in an easy way, not too technical and step by step so create some documentation and tutorials as well internally.

Okay, there is a question about do you find Google BigQuery to be scalable and budget friendly compared to other options like AWS?

I've never worked with AWS so I don't know really. Of course, whenever I speak to my customers a bit about BigQuery, I say that it's budget friendly and it's scalable because I think it is. But I didn't work with Amazon tools to compare.

There is another question as well. What data warehouse is best for security? I don't know if that’s a question about for the security data to store your security data and analyze if there are some issues with security. I don't know if there is the best data warehouse for that. I think all data warehouses that have data protection are good for that.

And what data warehouse is best for security if you ask in terms of whether they are secure enough. I would say the most known data warehouses on the market actually pay a huge attention to security. There's always a page that explains different compliances and certifications, and how it stores data and how it secures data. So it's always quite well explained in the documentation.

Okay. I think we will need to wrap up for now. Go ahead and add me on LinkedIn if you want to ask me questions there.

Turn your marketing data into opportunity

We streamline your marketing data so you can focus on the insights.

Book Demo