Click to learn more about Dave Karow.
What good is data without context? Most teams have multiple sources of data that capture customer behavior and application performance, but few have connected the dots between that data and individual code changes. Watching for a spike in a graph after doing a deployment may be “easy,” but it’s neither smart nor scalable.
Tools like Google Analytics, Sentry, New Relic, Segment, and mParticle are used widely and are more than capable of tracking app performance and customer experience. These tools can be especially helpful when rolling out new features, but until recently, it has not been possible to integrate them easily with feature flag platforms.
Feature flags are used by developers to toggle system behavior without changing the code. Feature flags separate code deployment from release in order to enable continuous feature delivery at scale. Development teams deploy code frequently, release features when they’re ready, and use data to ensure the release is successful.
Integration between web and customer data platforms and feature flag platforms delivers many benefits to engineering teams. By pairing web performance data with feature flag data, these teams have a direct understanding of how key service-level objectives — such as page load time or errors per session — are impacted by each new feature they release. These integrations can also drive alerts and notifications based on releases or code changes, making it fast and easy for DevOps teams to identify what caused an issue. Also, by using existing customer data pulled from these platforms, enterprises can run full-stack experiments and A/B tests that prove which features perform best for both the customers and the business. This allows teams to automatically collect metrics instantly as a user is browsing their website or using their app.
As an example, Segment allows for the centralization, management, and activation of customer and engagement data across any channel. It is a vital tool for all line of business and product people already, but with the right integration, it becomes equally valuable to DevOps practitioners. Now, due to tighter integration to some feature delivery platforms, Segment can show DevOps teams essential information like the average revenue per customer or the number of support ticket changes in response to application changes. Additionally, it works well in reverse by forwarding a record of every feature flag evaluation for use in a data warehouse or analytics tool, enriching customer data with a history of each feature a user has seen.
Google Analytics (GA) is already in use by 86 percent of top websites to report on web analytics, including performance and business metrics. This is another example of a tool that is used mostly by line of business and product folks but full of value to DevOps teams once they can filter it to focus on specific feature flag rollouts. That filtering or overlaying of data becomes a trivial operation once your organization implements an integration between GA and your feature flagging platform. Again, this is about the value of context. Imagine being able to observe the impact of every feature shipped to customers using web performance data you are already collecting in Google Analytics.
With these critical integrations beginning to rise in DevOps, teams are closing the feedback loop from the features built and delivering quantifiable customer and business value faster.