I led product design at Datagran, a low-code platform for data scientists, analysts, and business analysts. My role included overseeing the entire design process, from research to QA, and I collaborated with stakeholders and developers to make product decisions.
My main goal was to increase customer satisfaction, and I managed to raise the NPS score from negative to 20 positive in one year.
Some of the features we developed and launched included a churn monitoring and optimization feature, which used connected data to predict audiences that were close to churn, and to send custom campaigns with micro-segmented audiences. We also created a budget allocation tool that used AI to distribute a budget across multiple campaigns, and a business intelligence dashboard that could import transactional data, creating segments to target on Facebook and Google.
We regularly test and validate our product through a variety of means, including user interviews conducted with individuals from around the globe. In addition, we employ internal surveys to measure customer satisfaction and gauge the usage of certain functionalities.
Research revealed that data scientists and analysts often struggled to bring their work into production and make it meaningful for the business.
This is why Datagran was developed: a powerful, collaborative tool for large organizations that enables faster collaboration and production.The process was straightforward.
First, users connected various data sources using our user-friendly wizard. Then they could create organizations, invite users, and create projects with their own custom dashboards. Once connected, they could build machine learning pipelines, data visualizations, or use any of our AI templates or reports.
Our research found customization to be a major discovery, so we created an SQL editor that allows users to write code or construct data visualizations with a graphical user interface. With our interface, users can customize their visualizations by selecting one or multiple data sources and filtering and picking the desired values. In addition, we have provided a graphical interface to easily aggregate and filter data.
We have created a user-friendly visual interface that enables users to construct machine learning pipelines quickly and easily. Through a canvas and a selection of flow creation interactions, users can select data sources, aggregators, and actions before finalizing their output.
Datagran it's a complex product that involves Big Data, AI and Data Science. The challenge was to design solutions for user that are complex but usable, allowing people to interprate their own data and obtein powerful insights