The client was a private equity backed SaaS company that provides remote access software to businesses. Their product range includes desktop access and IT support across various pricing tiers.
As with most SaaS businesses, our client had built up an extensive amount of data about their products and customers but crucially lacked a consolidated view of the key functions of the business, namely marketing, sales, and product usage. This view is an essential building block to sophisticated business management in the digital world – understanding the behaviour of a business’s customers today in order to predict what they might do tomorrow.
Data cubes are the means to unearthing those behavioural patterns and delivering that consolidated view. This in turn drives business decisions and commercial impact, such as reallocating marketing spend to where the highest value customers can be found. Equally, they strengthen future valuations as a result of the improved governance and insight that high quality analytics affords. Finally, the cubes provide a strong foundation for the client to become a M&A player within their field, making data integration from buy-and-build acquisitions cheaper and easier to complete.
The client therefore had two goals for this project:
Democratise their data supply and make it accessible to teams across the business. In other words, if a manager had a hunch about a new initiative to drive sales, they should be able to access the data that proves their hypothesis.
Understand their customer behaviour in greater detail in order to improve customer acquisition, prevent customer churn, and increase market share
To support that vision QuantSpark was engaged to build three data cubes – streamlined models of datasets with defined use cases. These included:
Subscriptions and contracts - focused on surfacing key financial information for board and operational reporting.
Product usage - to identify and predict the indicators of customer churn
Marketing attribution - used simply to identify which were the most profitable customers and which marketing initiatives were most effective at driving them to the business.
Done effectively, the cubes would directly impact revenues by helping the client to both identify high value customers as they entered the sales funnel and highlight customers most at risk of attrition, to say nothing of the man hours saved by automating detailed board report creation that had previously been built from scratch each month.
QuantSpark’s approach to building data cubes can broadly fit into several key phases:
Discovery - where we assess the client’s existing data infrastructure alongside the quantity and complexity of the data itself.
Data modelling - cleaning, arranging and structuring the key data sets to make them easily accessible to the client’s analysts.
Data engineering - building out the plumbing systems that help data to flow, as well as establishing a cost-effective cloud storage environment.
Visualisation - creating a suite of dashboards to display the client’s desired metrics and information, achieving their goal of ‘self-service’ analytics for their business teams.
Our first step therefore was to assess the client’s existing data engineering that would form the foundation of the project. The client’s pipelines were robust and more than capable of handling the quantities of data flowing through the business, but of course the data itself was stored in obscure non-relational databases.
Consequently, the logical next step was to ‘relationalise’ that data, extracting and aggregating it into manageable volumes that could be interpreted and used to inform business decisions. From here we supplemented the process with an additional pipeline to connect to key marketing information in Google Ads and Google Analytics – key sources to understand where high value customers fell into the client’s sales funnel.
With a full database, we connected DBT, an open-source SQL query tool that facilitates an easy and cost-effective build of the data cubes, minimising engineering requirements that would have increased project costs. Not only would this make the Visualisation phase easier, but DBT also meant a cleaner, readable data model that the client’s teams could analyse and maintain.
Through these cubes and effective back-end data pipelines, we achieved the client’s goal of democratising data to increase their teams’ autonomy and decision making across the business. We were able to automate the client’s board reporting process, saving significant man hours and reducing key-person risk by ensuring their knowledge now lived within the business.
While the benefits of data engineering are often not immediately clear to a business user, as each cube has a specific use case – understanding their customer purchasing behaviour and where to find high-value customers – it creates a clear business case for ROI to be measured over the coming months.
We fully expect the financial return on investment to grow over time as they are able to identify more and more top-line high value customers through our new marketing attribution cube, and utilize the product usage cube to find the behaviours that define when an existing customer is most at risk of cancelling their subscription and leaving the business, improving the client’s bottom line performance.