Setting up an analytics infrastructure, one step at a time
Agilis Systems provides GPS fleet tracking and management for industries such as trucking, construction, and equipment rental. Agilis Data Scientist Stephanie Nguyen's job is to translate her company's data sets into meaningful data insights, but she faced a number of roadblocks. "We wanted to focus on analytics, but there was no data infrastructure in place to combine data from different sources."
Setting up all the components of a robust analytics infrastructure takes some time. Nguyen says, "I took about three months to set up the PostgreSQL database for our data warehousing infrastructure. That included mapping our business processes in the production database to understand the data we needed to ingest. Once I understood how production works and our requirements for reporting in phase one, I set up a denormalized PostgreSQL data warehouse to receive the deposited data.
"We also needed an easy-to-leverage ETL tool that didn't require a lot of engineering on our side. We looked into half a dozen solutions. We picked Stitch because it required the least amount of development knowledge, and was easy to set up. Many of the others required too much engineering knowledge to implement, or were too pricey."
Nguyen got Agilis set up on Stitch's free tier, which lets anyone replicate five million rows of data per month for no cost. "That allowed me to experiment and figure out how these integrations worked from a data analytics perspective, without having to ask for money from the company before I had a tangible proof of concept to justify the expense.
"We used Stitch to integrate our advertising data from Google AdWords into our data warehouse. That lets our marketing team join sales data with marketing expenses to gather insights on campaign performances. Stitch was easy to set up; I spent more time analyzing and understanding the data than I spent creating the integrations.
"Previously, we generated a manually created report every month that took about 16 hours to prepare to provide these marketing insights; now we have the information automated on a daily refresh. That both saves us time and gives us heightened trust in the data, since we're consistently pulling the same data through the same rules."
It's great to be able to set up integrations easily through the interface, but it is also great to have the documentation knowledge base and the extraction logs and error messages Stitch provides to help me troubleshoot problems quickly.
Nguyen says the amount of information Stitch provides data scientists and engineers is one of its strong points. "As someone working in the range between pure data analytics and pure engineering, it's great to be able to set up integrations easily through the interface, but it is also great to have the documentation knowledge base and the extraction logs and error messages Stitch provides to help me troubleshoot problems quickly.
"Also, the new free historical data load policy for new integrations is fantastic. It makes me less fearful of making mistakes during the initial testing of new integrations; row count costs can be expensive for a small company!"
"The data analytics ethos at Agilis is still very new," Nguyen says. "We're slowly accumulating data sources." She says Agilis plans to integrate data from Quickbooks Online, Salesforce CRM, Salesforce Marketing Cloud, and Google Analytics, and leverage Stitch's Import API for several additional data sources. "Once we're up and running with them all, we'll start adding more row allocations and move into a paid plan."