Here’s how we upgraded our marketing analytics

I hate interrupting my analysis workflow by tabbing between different applications and interfaces. It’s irritating, decreases your productivity and just makes things harder to understand. Therefore, I could empathize when one of our marketing people came up to me and expressed their need for an online marketing dashboard. In their vision, this dashboard would unite all our most important online marketing indicators and help them immensely by removing the need to go back and forth between the analytics views of different platforms.

But online marketing data is isolated, lives in silos and the individual platforms don’t make it easy to integrate them with one-another. Luckily, most of them offer API services, so we rolled our sleeves up and built a basic data pipeline, which resides entirely in the cloud and feeds our Tableau dashboard.

The data

As far as social media platforms go, Starschema mostly uses Facebook and, to a much lesser extent, Instagram and Twitter. Our leads are generated through our website, the traffic of which we measure with Google Analytics, which we also use for our standalone, Wordpress-based blogs. It would have been nice to get the traffic data from Medium as well, but this platform doesn’t offer an API for that unfortunately, so it’s not currently in our scope.

The pipeline

If you’re only interested in the visuals and not how the data got there, just skip this section.

We are not a small firm anymore, so whatever we would have created needed to be as enterprise-ready as possible, not least because we wanted to showcase this and use it as a proof of concept for other projects. We also wanted something with low cost and maintenance, since we want to be able to deploy this for smaller firms that might not necessarily have tech personnel on board.

Thus, we opted for the Google Cloud Platform, mostly because their generous free tier ended up completely covering our requirements. The idea is to have scheduled Python scripts download the data through the API-s, flatten it and load it into our Marketing Data Warehouse which we set up in BigQuery for the sake of simplicity. In a more mature environment, we would put a frontend onto App Engine to drive the Scheduler and the Functions, but in our case we skipped this and manage everything through the GCP console.

The very simple pipeline architecture we set up for this project

The dashboard

Our first iteration of the dashboard

Using Tableau for the data visualization bit was a no-brainer. First of all, our entire internal business intelligence suite is built on it which would be enough reason on its own. What’s more is that it’s the best BI software for easy visualization as long as you have the data nicely in order. And since our data pipeline is very straightforward and the data is easy to work with, we didn’t need to look any further.

The first iteration of the dashboard has two goals. On the one hand, we want to get an overall picture of the results of our marketing efforts. On the other, we want people to take it as a starting point and start asking their own questions. This duality is reflected in the layout.

The overview section allows us to see in two seconds whether we’re doing fine. The top set of indicators is about the quantity, while the bottom set is about the quality of the traffic. You can see three different chart types on the bottom of the dashboard, serving as an appetizer for our marketing peeps, so they can start thinking about further things they’d like to see, more complex questions they’d want to ask. The horizontal split, in turn, is based on platforms, packing our charts into a neat little grid, so we can keep the layout tight.

Yay, marketing analytics is done for Starschema! … or is it?

Of course not. Analytics is a process, not a project and there are no exceptions. As I said earlier, we want our marketing team to be hungry for insights. We want them to come up with questions but more importantly to be able to answer those in a self-service fashion. As the famous saying goes “Give a marketing manager a dashboard and they’ll understand things today, teach them self-service analytics and they will derive insights forever.

So, we’re moving forward by iterating the content and design of this dashboard as well as coming up with new ones and we don’t expect that this process should ever stop, we just want them to start doing it for themselves.

Is this something your company would benefit from as well? Do you think we could help in setting up marketing analytics for your organization? Get in touch!