CircleCI Enables their Product Team to Measure ROI by Activating dbt Models
While the adage of “move fast and break things” exists, what if you could move fast but not break your product in the process? That's exactly what CircleCI does by enabling developers to experiment, deploy and test code extremely fast. You no longer have to worry about a tradeoff between speed and reliability: CircleCI provides you with both as a leading CI/CD solution.
Results
- CircleCI has decreased sync time to Salesforce from 20 hours (with in-house tools) to just minutes with Hightouch.
- CircleCI uses Hightouch to sync their qualified users data model directly to Amplitude and determine which behaviors are most likely to translate into future customers
- Syncing product usage data and key financial data to Salesforce has enabled CircleCI to identify product-qualified leads and analyze contract utilization
- Using Hightouch, CircleCI syncs machine learning models directly to Gainsight to predict and reduce customer churn.
The Problem
As one of the fastest-growing companies in America, CircleCI has experienced record-breaking growth since its inception in 2013. This growth created problems for CircleCI though because as the number of users and data sets increased, data access for business users began to take longer and longer.
We were having scaling issues. We needed to speed up our existing data processes to unlock access to that data for our new users. We had some jobs that would take almost 24 hours to update a day’s worth of data and we couldn’t keep up.
Seamus McKinsey
Senior Analytics Engineer
•
CircleCI
In order to facilitate this growth and address the scaling problems it was facing, CircleCI was forced to move away from its legacy Postgres data warehouse and adopt a modern technology stack consisting of Snowflake, Airflow, dbt Cloud and Looker.
This quickly solved CircleCI’s problem around scalability and speed but also created another problem: all of their data was locked in Snowflake and only being used for BI reports. CircleCI needed a way to move data from Snowflake to Salesforce, Amplitude, and Gainsight to empower their Sales, Product, and Customer Success teams.
Using a home-grown data pipeline, sending data to Salesforce could take up to 20 hours.
Our Salesforce data pipeline was the bane of our existence. It broke on a weekly basis, took 20 hours to complete, and there was no easy way to view what data had been updated without looking at the logs. Adding a new field would take several hours because we would have to test it and figure out why it failed. With Hightouch, we can do all of this in seconds.
Seamus McKinsey
Senior Analytics Engineer
•
CircleCI
The Solution
In search of a solution to eliminate this problem, CircleCI quickly discovered Hightouch.
Developer features like a Git integration ,dbt Cloud integration, a Live Debugger, and a large catalog of integrations were key factors in our decision to adopt Hightouch for Reverse ETL
Seamus McKinsey
Senior Analytics Engineer
•
CircleCI
Empowering Product teams to identify drop off and measure ROI of experiments
With over 800 thousand developers and 30 thousand organizations using the product, CircleCI generates tons of new users on a daily basis. At this volume, it has become a challenge to determine which leads have the highest buying intent and proactively target them. While CircleCI’s Product, Engineering, and Marketing teams use Amplitude to understand user behavior, they needed a way to run & analyze experiments on qualified users, not just signups.
This proved to be challenging because CircleCI’s data model for “qualified users” only existed with Snowflake (modeled in dbt). Using Hightouch , CircleCI syncs this data model back into Amplitude so that these various teams can create KPIs and track experiments for new features and campaigns to answer questions like:
- Where are our qualified users dropping off?
- How can we design a more intuitive flow to increase conversion for qualified users?
With Hightouch it was really simple to set up a dbt model around a set of facts that we wanted to sync back to Amplitude as events. We just have to make sure the logic is correct and Hightouch takes care of the rest.
Seamus McKinsey
Senior Analytics Engineer
•
CircleCI
Empowering the sales team to identify PQLs
CircleCI uses Hightouch to sync new users into Salesforce and enrich those users with data on how they are using the product and how much they are paying. The sales team then uses this user-level information to identify product-qualified leads (PQLs), upsell opportunities, customer health scores, and analyze contract utilization.
Hightouch has decreased the time it takes to sync data to Salesforce from hours to minutes compared to CircleCI's previous in-house scripts. Making changes to pipelines is much easier now: it used to take CircleCI multiple hours to add new fields, but with Hightouch, new fields are added in minutes.
Before Hightouch, CircleCI used a number of custom integrations with other financial tools like Stripe and Chargify to send revenue metrics (ex: MRR or Monthly Recurring Revenue) directly to Salesforce. Because of the point-to-point nature of these custom integrations, they were error-prone, leading to inconsistent metrics and data reliability problems. Business users did not trust these metrics. With Hightouch, all of these custom integrations have been eliminated and all of this data can now be pulled directly from a single, reliable source of truth: Snowflake.
Empowering Customer Success teams to identify and reduce churn
Customer health is an important indicator for CircleCI’s Customer Success team, so the data team created a machine learning model to predict customer churn. Hightouch syncs this model to Gainsight to flag customers who are at risk and might require further actions and trigger specific workflows.
Before Hightouch our data was not up to date and not reliable. Now everyone trusts the data because everything is highly standardized and documented.
Seamus McKinsey
Senior Analytics Engineer
•
CircleCI