How did I start Growth Engineering at ShiftLeft?
How a growth analytics infrastructure was built in 1 month. This is part-1 of two part series.
ShiftLeft became a product-led company in 2020 with the introduction of self-serve onboarding. With self serve, users are able to sign up for the product, use it with demo apps or their own apps, invite their team, self-upgrade to a premium trial, and try out integrations and APIs without interacting with a single person from ShiftLeft. That was great for users discovering our product, but now we had a problem: we need to know what our users are doing.
We didn’t have the resources to retrospectively develop instrumentation within the product. We already had released self serve and we realized that instrumentation will only generate events from the time of implementation, so it wouldn’t tell us what existing self serve users did in the past. We had to use data we already had. And that data was in our product database, Redshift, and HubSpot.
Visualization using Amazon QuickSight
We needed a way to use visualize product usage data from our sources of truth (product database, CRM, warehouse, etc.). Amazon QuickSight was the BI product we decided to use. Our infrastructure was already in AWS so it was the most straightforward service to add among the BI options we evaluated. It also helped that using Amazon QuickSight helped us with our SoC2 compliance because we didn’t have to introduce a new vendor.
Modeling accounts
Amazon QuickSight is a powerful BI product but we still had to figure out how to use our product usage data. Our main objective was to know how users and accounts were progressing through our usage funnel.
The numbers represent the number of accounts at each stage. Knowing that end result, we worked backwards to define an accounts model. This data modeling approach was the key driver for the rest of the growth analytics implementation.
The accounts model is a simple, high-level representation of each customer account in our product. It’s defined in QuickSight using a SQL query that fetches each account and all of the properties and aggregate metrics we’re interested in.
Using this modeling approach, we were able to slice and dice the data, create different visualizations and tables, and use filters to understand our self-serve growth from several dimensions.
Part 2 in this series will go into more details about how workflows were created to engage with users using product usage data.