Know if Your Features Cause Good or Bad Outcomes
Split’s patented Attribution Engine joins feature flag data with performance and behavioral data to measure the impact of every change you make to your app.
This feature-level observability allows Split to catch every unexpected consequence of your rollouts quickly, even when issues are small. We call this capability “causal analysis,” and it serves as the foundation for all of our measurement and learning tools.
Catch Issues Automatically, for Every Rollout
Set up your organizational metrics once and let us take care of the rest. Once configured, Split automatically calculates metrics for every future rollout. Whether you’re rolling out three features a day, or thirty per hour, Split will catch any issue.
Are you doing complex releases with many features? Split will identify exactly which variation of a feature flag is causing unexpected consequences and send an alert to the right team, so that you can quickly take action. No manual triage work or war room needed.
Moving Fast at Scale Takes an Integrated Solution
When using simple ON/OFF feature toggles, engineering teams typically rely on time-based correlation to detect issues. But, by the time you catch issues, they’ve already caused big problems for the end user. That’s neither safe nor efficient.
Today’s teams require an integrated and automated approach. Split seamlessly pairs gradual rollouts with automated monitoring, measuring the impact on any metric you care about. With limited exposure, early detection, and the certainty of causal analysis, your rollouts leave nothing to chance.
Find & Fix Issues Your APM Will Miss
Your APM tool is essential for monitoring at the application and infrastructure level, but it will only catch an issue when it becomes severe enough to rise above the overall noise of the system.
With Split’s feature observability, the impact of every change can be monitored more precisely to detect small issues during your rollouts way before your APM tool.
Consider a situation where 2% of your customers are getting an error. That’s unlikely to stand out enough to get noticed above the noise in your APM tool. In that same situation Split knows that the two errors are in the new version of one specific feature. From that perspective, 40% of those users are getting an error. Seeing that, Split will fire an alert, offering the right team an immediate and targeted way back to safety.
How Split Compares to Your APM Tool
The goal with APM is detection. When things go well, the impact is a reduction in mean time to detect (MTTD). Once you know something is wrong, triage is manual, which means it’s up to you to find a resolution. Mean time to resolution (MTTR) can be hours or days.
With Split, the goals are detection, triage, and resolution for even faster MTTD and instant MTTR. You automatically have fewer blind-spots and can say goodbye to long-running incident rooms that divert resources from getting new work done.
Split provides our team with rich data on every feature, allowing us to make accurate changes in our application.”
Miller Dugalech, Director of Digital Management, Quility