This content, written by Áine Dundas, was initially posted in Looker Blog on Feb 24, 2020. The content is subject to limited support.
In this blog, we talk with Alexandra Sudilovski, Head of BI at Appsflyer. Appsflyer is a rapidly growing SaaS company, and we wanted to learn more about how access to data has helped to fuel that growth.
Hi Alexandra! Could you share a little about your background in data and your current role in Appsflyer?
I am a BI expert at — a mobile attribution and marketing analytics platform that helps application marketing teams around the world make better decisions. We enable marketers to measure everything from impressions and clicks, to installs, purchase activity, ad revenue, and so much more.
One of our main responsibilities as a BI team is to provide data and make it relevant, available and accessible to all our internal departments. That’s no mean feat! I’ve been working with Looker for about two years already, and I’m focused on really integrating analytics into our teams’ day-to-day workflows.
That sounds like A LOT of data to be processed!
It is indeed. We gather and collect all the information that is coming up from mobile advertisements and applications; collecting it and saving it in our internal databases.
Our product is a nice sandbox for marketers, that contains a lot of extremely important KPIs for their business. We have all of the performance metrics — marketing KPIs such as the cost, clicks, the CPM, CRM, and ROI.
So if we try to quantify that — just how much data are we talking about here?
We are processing about 80 billion events on a daily basis.
This is BIG data. And this is at the aggregated level. At the raw‑data level, we have about 70 terabytes of AWS data every single day. And in BigQuery, we have around 27 petabytes of data — that is raw data events that have been stored in the last three years. And as Appsflyer is continuing to grow very fast, you can imagine how much data we will have after every year of work.
We have a lot of departments internally. They all want analytics. They want to make better decisions so that we continue to grow as a company. As the BI team, we want to meet these demands, better inform users across our company and fuel growth.
That’s certainly the experience many organisations have. The appetite for data — across multiple business functions — is something that is constantly growing. What was your starting point in evaluating how to meet these demands?
I evaluated several BI tools, including Tableau, Sisense and Looker. We decided to go with Looker, as it immediately gave us all the solutions to our needs. It’s a real‑time analytics platform, meaning I'm connecting all my data sources directly to Looker without any data warehouse or any processing in between. When I'm running the query or sending the query, it's going to the database and giving me results. So, I really have the advantage here of the engine processing of Google BigQuery. This is a huge database with a huge potential. Also, all our cat files are in Athena, which also can make it very quick to analyze vast amounts of data.
Looker has a modeling layer, which is based on LookML. It’s a SQL‑based language that is really intuitive, making it quick and easy to learn. You can build on top of your own data or aggregated data. It can really build a business logic, which is extremely important for different departments.
In terms of performance, it is that real‑time connection that allows us to take full advantage of our data sources. The performance is great — and has lots of possibilities including the caching mechanism or defining triggers which will send the query only if that data was changed.
For permissioning, we have several types of users who have different access levels. You can be a viewer. You can be an editor and a content creator. You can be a developer who is really working with LookML, and building all this business logic layer. And, of course, we have our BI team. We are administrators — we're providing all the connections, all the permissions.
You mentioned that you have a great demand for access to data across various business functions. Do the permissioning levels you talk about cater to those internal ‘personas’ who want to leverage data?
Absolutely. Once we started to roll-out and work with Looker, we saw extremely great engagement. We started to train all our internal departments in a phased approach and bring them to Looker one by one. So, we started with three departments — customer success, sales, and product analytics. That was over two years ago.
Now we have almost every internal department working with Looker — including finance, support, learning and development, and HR. To support these various business needs, we created internal documentation and training. In fact, in our onboarding training for new employees, we cover what Looker is and how to use it. We have customised training options for various users. This differs for viewers, versus those who really want to dig into the data and create their own visuals and dashboards. To be a developer, you have to learn LookML.
Sounds like you had a strong vision and an impressive roll-out that really drove adoption in the company.
The adoption rate is really amazing. We currently have 50 analysts working with Looker. We have 560 users. More than half of them are explorers — drilling into the data and leveraging it to create visuals and dashboards. They have more than 500 dashboards. Currently, over 75% of the Appsflyer team are using Looker — and I see the numbers going up every day.
Ultimately, our vision as the BI team at Appsflyer is to provide self-serve BI. And this helps to deliver on that. Appsflyer is a very data-driven company. We have analysts in every department, to deeply support that department and understand precisely what are the most important KPIs for that department.
Discover more about how Alexandra and the team are fuelling growth at Appsflyer in this .