OWOX BI Research on the State of Digital Analytics: Interview with Anjana Aggarwal from Hubspot

Digital analysts are always busy. Especially if they’re Anjana Aggarwal from Hubspot. We’re really grateful to her for answering Marіia Bocheva’s questions as part of OWOX BI research on digital marketing analytics.

Here’s the list of questions for you to navigate:

Skills block:

• What hard skills are most important for analysts today?
• What soft skills should a good analyst have?
• Does an analyst have to know SQL, Python, and R and build compiled dashboards?
• What’s the biggest mistake an analyst can make?
• Can you share some of your analytical mistakes?

Communication block:

• Do you think miscommunication between analysts and marketing teams is common?
• What knowledge are analysts and marketing specialists missing in order to make companies data-driven?
• How can an analyst have a greater impact on marketing? How can they be useful for the marketing team?

Challenges block:

• What are the most important things analysts need to do at different stages of business maturity (startup, SMB, SME, enterprise)?
• What analytical challenges do you have at your company right now? What tools do you need to overcome them?
• What difficulties do you see when it comes to implementing analytics and how would you assess the overall development of the market?
• How do you evaluate the current maturity of marketing analytics?

OWOX BI conclusions

Anjana Aggarwal: I’d say it’s good to know statistics and SQL. I think these two are the most required skills. Also, knowledge of Excel is important. Actually, there are many other tools showing up every day, but you can’t learn everything. Basically, if you have a good analytics eye you can learn how to use new tools easily – for example, Google Data Studio.

If you have statistics knowledge and you can make sense of data, you can use any tool.

AA: First of all, it’s communication skills and the ability to make your data easy to understand. When you work in statistics, you get used to statistical language, but sometimes other people you communicate with can’t understand it. This also means that you have to use the skill of telling a good story and making sense of data.

AA: SQL is important for analysts, R for statisticians, and Python for data scientists. But there is no specialist who uses everything at the same time. This is not possible. In the company, you shouldn’t mix up those roles.

Basically, R comes in handy. Knowing R is good, but it’s not a rule and you can’t reject a person who doesn’t know R because this is a new thing. And if you are a statistician, you can learn it easily.

What about Python? I feel that Python is not for data analysts but for data scientists. If you are hiring a person for this role, knowledge of Python is important.

AA: You should always think about that. If you can present it so it makes sense, okay, do it, but in one graph only. That’s how it works with people in business: you cannot expect them to look at five graphs at once.

Knowing how to present your dashboard is also very important. I used to make some mistakes with it when I started. I used to see dashboards only from my side, and then I got many questions from top management. And I got the feedback: this doesn’t make any sense.

So I went to people and asked them about their concerns. I asked them, What data are you looking for? How do you want to see it? How do you want to use it? How could it make more sense for you? Yes, graphs are not new for them; they use them in Excel, for example, but they sometimes have another vision of visualization. So I accepted my mistake, asked them questions, and asked them for their graphs to understand how they see them. And it helped me improve my charts. I improved my vision and people could make sense of my data.

AA: Assuming that he is totally correct and not exploring further. Not going deeper and making assumptions about the data. My own mistakes were about implementing events incorrectly. And they always lead to spike of data value by several times. I didn’t know that it was incorrect. Then I went to Google Analytics (GA) and started exploring deeper and found out that something was not good. Also, I didn’t pay attention to concerns.

I worked with a multinational team who used GA with international properties and after some time we found out that they made much fewer assumptions, around 50% fewer compared to our system. They said that something was incorrect, but I didn’t agree because I was so biased. I said our system was absolutely correct and it was not reporting anything wrong. But after three to five back-and-forth emails I went to GA and, to my dismay, found out that there were some internal mistakes in the Google management console we maintained in consumer tracking. We didn’t see a mistake and implemented all as it was.

AA: Yes, absolutely. It’s a very common problem. For example, marketers have some statistics knowledge, but as they use other tools and metrics they can misunderstand or misinterpret the data.

AA: I think to overcome miscommunication issues, you should firstly make everyone around you understand marketing BI, explain the meaning of your metrics, and show how they can help. IT analytics is pretty new, and not many people understand how it helps them. That’s the main reason for miscommunication. And you should remove this block. You should talk to marketers around you to make them understand how they can use your data in business. In my job, I always face this challenge. I just step up and talk to them.

AA: Firstly, they shouldn't make any assumptions about the data. They should segment the data and go further down. Without segmentation, you can’t tell a good and true story with your data. They can think they’ve faced that kind of data before and make assumptions based on previous experience, but whenever they had such data before it could not work in other particular cases.

Every time they should try to use their analytical eye to see what is happening with the data. I think they don’t need any hard skills (SQL, etc.) for that, but they should analyze the previous data, verify it, and explore more and more.

AA: In short, analysts can help marketers find new opportunities, trends, and growth zones. Analysts can help businesses to understand what metrics are best to track. You see numbers and you can tell marketers, “this is doing the best,” “you should use this content.” You should optimize it. Maybe you can see an opportunity in the mobile phone or desktop. You see where traffic is better and suggest where it is better to invest. Or you can see in which regions a product works good or not really well and suggest how to spread the market.

AA: For a startup company, the most important thing is to use affordable, mostly free tools. Startups usually don’t have enough money. As an analyst, you should recommend which free tools they can use and how they can make sense of the data.

In return, startups should understand which data they need and which data they have. This will help them not to buy all recommended tools but to understand how to use free versions the most profitable way. Startup companies always need something to improve, and for you as a data analyst it makes for plenty of challenges, so you’ll always have enough work.

I think that for small and medium-sized business it looks almost the same. But you must understand that as a data analyst, you’ll have less data to work with but the number of data sources will increase. It’s very important for small, medium-sized, and even enterprise businesses to integrate data sources.

As a data analyst, you should recommend a process for how to operate with a large number of sources and how to gather data in one place so you can make sense of it. This is the main problem at medium-sized business stage.

What about the enterprise stage? I think they already have a pretty huge amount of data and you can use predictive analytics based on the big data from previous sources. Basically, you can use it everywhere, but at this stage it works the best.

AA: Right now we are struggling with combing too much data from different sources. We use different systems to gather different data because no source can give you comprehensive data. We use so many tools such as Facebook, Google Ads, and Google Analytics. We also get data from different applications with Firebase. If we want to collect data about customers and their behavior, we use Looker. And then to join this data we use Hubspot CRM.

We are also going to use SuperMetrics soon. Joining it all together is not an easy task. So the greatest challenge we are facing right now is combining data from a huge number of sources.

Note: Here, at OWOX BI, we know about the difficulties of data collection and merging a lot. But even in comparison to the most promising tools we’re still the best choice for companies with multi-channeled marketing:

• We know the specification of the collected marketing data. It helps us merge faster and with less number of collisions.
• The collected comprehensive data can’t be harmed by sampling. We gather the unfiltered data in its’ springhead.
• The number of sources, volumes of data — are not limited at all. We can handle a project of any size.

AA: Sometimes, people implement analytics for their business but they don’t know how to make sense of the data. So the main problem, I would like to say lack of documentation and expertise. For example, if you want to implement Google Tag Manager, you should read Simo Ahava's blog and learn.

Using Google Analytics or BigQuery is not so easy if you aren’t a technical specialist. There is a lack of resources that explain how to use it in “human language.” You have to read much documentation to understand it. You are always floating around. There is no system with documentation you can read and understand everything. We spent too much time to understand how to implement any tool correctly. The process is not clear.

I have a good example of it – Google Optimize. We used the free version, and it has a basic script for implementation. It says that you have to put it on your page, but it doesn’t say that if you do it, your site will become slow. It heavily increases your load time. It was hard to find the best solution to implement it without slowing down loading. It’s hard to find the balance between implementation and UX.

AA: Basically, I evaluate it at the starting stages. At the starting stages businesses building marketing analytics should use their own defence systems, their own data hits, their own system where they download data manually and then make sense of it. There are so many data sources and if people are doing great with them, you can see that straightaway.

While reading this interview for the first time, we were asking ourselves if the analyst’s life could be described more honestly than what Anjana has offered us. Also, we agree that the documentation issue is extremely important for both service provider and client.

Have a look at the OWOX BI help center if you want to learn more about the technical aspects of how our service works.