Paid marketers would naturally rank data management as a top priority to get right. However, many UA managers may not be getting their data in check. As many as 84% of CEOs are concerned about the quality of data they’re basing decisions on. At the operational level, it’s critical to present and manage data so all stakeholders can have a good grasp of what’s going on.
Tale About Data’s Managing Partner, Lior Barak, addressing a packed Mobile Spree audience, walked us through real data management. To be an effective app marketer you have to eat data for breakfast — but Lior took this concept to a whole new level, showing how data and hummus both have to handled with care. Through this simple analogy, Lior was able to empower Spree attendees with the steps a successful mobile marketer takes to translate CSVs into fully-fledged, automated reports. We’ve got the full transcript, video of the session, and essentials for you to catch up below.
Box of chocolates: Borrowing from Forrest Gump, Lior says, “data is like a plate of hummus.” Hummus is hard to make, and “it never looks good,” but “it tastes good” — and it nourishes you, potentially for a hard day of work ahead. “The idea of data is that we need to consume it in the right way and we need to be able to actually process it correctly,” much like the dish. If critical components are wrong, you could ruin your meal.
The daily grind: Lior suggests having three tiers of data. Tier 1 is for raw data, which should be accessed by engineers and Data scientists. Tier 2 is for marketers and analysts who wish to research their data. Tier 3 is for decision makers in the company — simplified reporting to give easy and understandable action points to decision makers. This way, marketers can control access to the right data, while reducing the noise for people who don’t need to hear it.
Controlling your diet: Much like hummus, you shouldn’t consume too much data. “I'm making sure that I'm limiting it,” says Lior. “On a daily basis, I can ask up to three questions, when it's coming to a weekly I'm going to ask 10 questions and at a monthly view, another 10.” This way, you’re not overwhelmed with information, or binging on data without anything productive to show for it. Having daily performance helps to “spot almost immediately if there are any errors or any mistakes.” While the latter gives a broader view of the entire organization.
The full transcript
So what we're gonna talk today about is actually about hummus. I know that you had a lot of talks about data and information and how to process it and stuff like and that and let’s face it, it's too heavy.
So what we gonna talk now, is about Forrest Gump. If you remember him, he said that, "Life is like a box of chocolates" right? So I say data is like a plate of hummus. Basically the idea of data is that we need to consume it in the right way and we need to be able to actually process it correctly.
So it's the right side and what happened is that when we have a lot of plates on the table it's really hard to figure out what we need to take out of it and when we have data we should be more focused and then this hummus plate is actually coming from Israel. This is from Abu Hassan in Yafo if anybody was there.
When we consume data we should be able to actually take the plate and really enjoy our data and really make a good meal out of it. Because usually in Israel, when you eat it, it's serving you for the rest of the day and the same thing with the data. Once you receiving the data in the morning you need to be able to serve it for the rest of the day with you.
And then the question starts right now, okay what are the rules to have a good hummus plate? And as a hummus cooker I can tell you it's a very hard work, it's not that easy.
The first thing you need to do, is to get to know your data and every time that you're getting your databases, the first thing you need to do, is to start understanding basically what is it inside their. Can you use it? Can you trust the data? What is the source and is there any other source that says otherwise and you need to be able to own your data. You've paid for it, you need to understand what's going on in there.
And when you're looking at data, usually on a daily basis, you need to be able to spot almost immediately if there are any errors or any mistakes. The best way to do it, is by owning your data. Don't let you data engineer do it for you. Do it yourself. Start digging into your data and understand much better how to use it.
There are three pillars for using data. For consuming data we need to be able to separate it into three different circles.
The first circle is the raw data, this one will allow you basically to go to the rawest levels of data. This means the raw information that's coming from Facebook, raw information that coming from Adjust, raw information that coming from Google and so on and so on.
Usually, as I see the raw data should be accessible only, by people that are capable of dealing with the data, which are usually data engineers or data scientists. That you want them to actually process the data and understand it much better.
The second pillar that we talking about is aggregation. After you got all this raw information you're going to start to aggregate it and you need to be able to analyze it and use it to some purposes. And by basically building this controls over your data, you're making sure that only people who understand the data and own it and understand how to use the data, can actually use it.
The third level is the visualization. This level basically making sure that people that need to make decisions, get the easiest and simplest database that they can start using. You visualize it for them, you give them some report and they can start immediately working. They don't need to start digging into the data.
This is a very important process to understand that we have this three levels, because each of them has control over it that making sure that no leaking of data or wrong data gonna go out to the company to make decisions.
When we're looking at it, actually how it should look like? So this is, for example, a very simple example that I'm usually using. What I'm taking is, for example, Facebook and Google API's data and I'm aggregating it into a level that I'm saying this is the performance and this is the cost table.
And then the second layer that I'm gonna take, is for example, is Google Analytics and Adjust data and I'm gonna aggregate it into an attribution table and then the third one will be the BI or whatever other services you're using to get your events, they're gonna collect it into an events table. This will be your second layer. This is where you're running your analysis, when you're understanding better your information.
Then when its coming to actually serving it to your customers or to your consumers or whoever's gonna need to use it, to make decisions. You're gonna go to the third level, which is the marketing tables. This is the performance table, this is the finance tables, what we basically doing by creating this structure is that we have a very safe QA system.
Because what happen is, that people receiving data, where they supposed to and they get it in the way that they can consume it and understand it and we do not mix them with a lot of different calculations and stuff that they're not gonna understand how to use.
And then of course QA and controlling of your data. Many of us having this issues that we arriving and we starting using the data in the morning, hopefully. How many of you using Excel actually, still? How many of you don't have a database that contains all this information in one place? Quite a lot.
So what we're doing is, for example if we need to use Excel sheets or CSV files that we 're downloading from our partners, we can start doing very easy QA on this information. The first and most easiest thing is of course to go over the size of the file and understand if it's too different than what it was before and then we can start processing if there is an issue or not.
And of course then we can go to the number of rows and we can start comparing our reattribution tool information, the campaigns that come from there, against the campaigns that come from Facebook. Are we missing something there?
It's very important to do QA's on a daily basis and understand basically if there is something wrong in the data because we need to make easy decisions based on our data and then the second level will be basically to start taking your information and then start processing in a little bit more deeper, this will be once a quarter for example when you're comparing your automatic API's data or your automatic data that you collecting into your data warehouse and you start to compare it against your partner.
So you go into your panels or to whatever dashboards you've got there and you start to comparing the numbers, manually. It's very important to have QA over our data because if you don't have a good QA over our data, we won't be able to consume it and the same thing goes with hummus.
You need to do a good QA on your hummus otherwise I will be mad because I will see you eating from a box that you bought in the supermarket. This will insult me because it's not the same.
Then of course when we're coming to pick the KPIs we need to understand that data is very important, we need to reduce the amount of data that we processing, we need to make sure that we focus on decisions and actions and not on guessing and gut feeling.
When I'm talking about it, I'm talking about it many times with a lot of people and they're saying, yeah I need at least 15 KPIs to make a decision if this companies working or not and then two of the impressions and two of the clicks and then two of that and I'm saying why? You have an end goal, your end goal is conversion, you want to convert your clients to do something, some outcome and then what you need to understand is basically, okay, what are the decisions that I need to make based on my data.
When you're arriving in the morning you need to focus on KPIs that make decisions easier for you. And this means how much money I invested is it good decision is it actually in our control, most likely not because we have a very fixed budget in advanced right? But if I'm investing it in a certain campaign or another campaign and I need to realize which campaign I want to invest it, this is decision I need to make. I need to make sure that this decision gonna be simple and easy for me to do.
This is basically how I'm doing tables, so when I'm arriving to the third level and I want to actually start making decisions in an easy way, what KPIs am I gonna include inside it? This is where I'm starting. I'm starting basically by naming out all of my questions that I have on a daily basis on a weekly basis, on a monthly basis. I'm making sure that I'm limiting it, on a daily basis I can ask up to three questions, when it's coming to a weekly I'm gonna ask 10 questions and a monthly, another 10 questions.
The idea of it, is actually that we controlling the amount of data that we consuming on a daily base, on a weekly and then monthly and we making sure that we do not consume too much information. And then once I have my questions I'm starting to define what are the KPIs that I need to answer this questions?
Do I need to have installs? Do I need to have costs? What is the KPI that I'm gonna use in there to answer the questions? This is the second level and this thing I'm usually printing on a A1 or A0, I'm hanging it in my office or in other offices and once every quarter we going over it again. And we try to understand, okay, what we made a decision with last quarter is it still relevant for today? Are the questions still relevant for us? It could be that we are in a different place today and some of these questions are not relevant at all. And to be able to actually make these decisions we need to have very important questions that being asked on a daily basis and then we can readjust it.
And this one is basically, with the sticky notes you can always remove them, you can always change them, this is the fun part about it. And then of course you're adding dimensions. So to what level do you need to go in each of the tables? So when I'm going to a marketing table or a performance table, what do I need to have there, do I need to have only campaign level do I need to have the country level, do I need to have an ad level? And then or maybe I want to see even the creative itself, and I need to understand how to do it and this is how I can actually make sure that I'm focusing on what is important for me.
When it's coming to reporting, and again like hummus, it never looks good, it tastes good. And this is what is important. We need to be focused on what we want to have and to be honest if you've been to Israel, you will know that always the shittiest looking, and I'm sorry for the word, but the shittiest looking restaurants will have the best hummus inside. And this is the secret, you need to go in the streets and understand exactly where you are and go into a restaurant that doesn't look so appealing but it's gonna give you what you want and the same thing goes with reporting.
In this report, over here, we're basically focusing on a very important part, for making decisions in the morning and it's basically, how is our campaign going? And should I change my bidding and if I should change my bidding, what is the percentage of change that I need to make? I have my call to actions that I'm using and then basically I'm giving myself an indication are they working well or not and should I change anything in there?
I see my breakdown of spending, so I understand exactly where my money goes to, this one has less like it because it doesn't say much, but my partner was forcing me to do it, so I did it, and then what is the driver of traffic to the website or to the app?
So to understand much better who are the best performers that we have right now, out there. And then of course what is the return of investment of this traffic because otherwise there is no reason to actually do anything.
When we looking at the reports we actually needs to start focusing on, as I see it, more control and less on only performance. We need to start focusing, on okay, how many changes did I do to my campaigns, when did I change them, how many did I change, was there any reason not to change some of them and I'm becoming more focused also on what I've done so I can remember it, because if somebody tomorrow needs to take up your place because you are sick, it's much more easier to have it in a very organized way.
You can download this data actually from any partner that you have today. All the logs of the changes and everything and you can just organize it inside a nice table. If you're looking at it one step further. If you're doing automation of it, then it's much more easier, because you know exactly what the machine was changing and what not. And the idea of having this one as a control is also for us to understand, are we going on the right direction or not.
While here we're understanding on the level of campaigns, are they working well or not? On here, we're understanding what are we doing exactly and we can always refocus on what we've done. We can always recreate this hummus recipe that is really good, right? If we have some log that actually saved it somewhere and the idea of it, is that we're becoming more data-friendly and we using more data to actually make decisions.
When we're talking about the new generations, where we're going right now, the direction that we're taking so there is a lot of changes in the market, we're moving into more automation. More ads controlled by Google, because they closing the garden. Right, UAC today, you almost have no control over it. Very soon maybe you're gonna have some but lets face it they're not gonna give you much control and this is where you need to start understanding that you need to be more data-centric, you need to understand much better your data and know how to control it much better and when we don't make decisions based on our data we have a big issue.
You cannot make decisions — 68% of the managers in 2017 said that they're making decisions based on gut feeling in the morning because they don't have their data in the morning. And this is a big issue. Cause lets face it when you arrive into the office you need to be able to make decision quite quickly and quite easily and you need to be very centric on your data and if you don't have it there, then there is an issue with it.
And then of course if you have data you can act fast, we don't want to wait for stuff to happen, right, we don't want to figure out that we spent thousands of Euros on a campaign that actually didn't work, but we didn't have data, so we kept spending money. This doesn't make sense. We should be able to work quite fast when we're making decisions.
And this is basically when we're looking at it, is to have the decisions made for us already in the reporting when we're already looking at the data. And then we can make our decisions much faster and much more easier for people to actually understand what they want to do.
And the most important thing, don't be lazy with your data. This is my hummus plate, this is what I'm making at home, usually once a month for some friends when they come over. If you gonna nourish your hummus recipe, you're gonna have a very good recipe to use. You gonna have a very good hummus, you will be able to sell it maybe to somebody, one day. But you will be able to actually make decisions quite easy and when you're dealing with data you need to be able to actually make your decisions easy, you need to be able to help people to make their decisions, easy and simple.
When you go into the finance people you need to be able to allow them to make decision on your budget, because they know exactly your performance. When you going to your manager, you need to explain to him, exactly what your performance are and how did you do it and this why you need to use your data, you need to get familiar with it, you need to use it, much more often than what you usually using your data.