Latitude Models
My Role
Founding Product Designer
Team
- Daniel Roig Software Engineer
- Gerard Clos CTO
Timeline
Q4 2023
Overview
Predefined SQL components to reuse and reduce the number of operations and blocks in the BI canvas.
Context
I worked as solo product designer in this project. We built a BI canvas based where each block is a SQL transformation. The problem was that use a clean dataset you had to repeat the same transformations in each project. We created models, a block that contains those repetitive SQL transformations allowing it to be reused anywhere with just one block.
Problem
Clean data working with SQL in most of BI tools is redundant and repetitive making loss a lot of time, generating mess and chaos.
In Latitude at the beginning was easier because of the canvas interface, but when the project grows, the number of blocks and operations increases, making it difficult to maintain and understand.
Research and constraints
Talking to users most of the time when starting a data exploration they had to mix different tables to have all the info from users, companies or other entities.
For example for users the company or subscription info was in different tables generating a lot of repeated joins and transformations.
Also any change in the definition of users would be neccesary to change in all the projects. So it didn't scale.
Solution
We implemented models, which is a block that contains those repetitive SQL transformations allowing it to be reused anywhere with just one block.
It's similar to a regular project but with an output block mandatory which is the result of the model or the definition.
This model then appears as source block in the canvas and can be used as any other block.
Success
The key metric here was reducing the number of source blocks from ~4 to 1. And the number of operations from ~6 to 1.