Harness Inc., the software delivery startup that provides tools for developers to update and monitor applications as they go, today announced several new features that expand its feature management and experimentation capabilities.

The new features come from the completion of the company’s integration of Split Software Inc., a company Harness acquired in May, into the platform.

Harness provides a continuous delivery-as-a-service platform using artificial intelligence, which manages code and software testing throughout the entire development lifecycle. Developers can monitor and manage release metrics, analytics and tests every time a change is made before they’re deployed to catch potential issues before they are released and roll them back if something goes wrong.

The Harness dashboard now goes beyond basic feature flags, which allow developers to turn on and off features with toggles in a dashboard. Now that Split is fully integrated, customers gain access to Harness Feature Management and Experimentation a back-end dashboard that fuses six new features into a centralized backend.

With feature management, users may now easily target specific users or segments of users to do A/B testing and roll out new features gradually. It provides robust permissioning, governance and control for enterprise organizations. Once releases are underway, FME can monitor for any degradation in customer experience, such as increased page load times, errors or crashes. It can automatically trigger rollbacks if certain thresholds are exceeded.

Using experimentation, FME allows users to run experiments by turning on new features only for select users to measure the impact on business metrics such as conversions, revenue or user acquisitions.

“When you look at the traditional Harness base you have your everyday continuous integration and delivery product,” Harness Vice President of Product Trevor Stuart told SiliconANGLE in an interview.  “That’s the bread and butter of the Harness feature management. Experimentation picks up where CD drops off. I get my code into production; I now want to gradually roll that out and be able to do it in a single platform.”

The experimentation feature goes even further to provide data-driven insights to optimize feature rollouts. This way developers and business users can have a complete view of how changes affected the performance of their app in real time, right down to user impact. It’s designed to ensure that every release is backed by empirical data to increase confidence in building a better product.

Experimentation now has its own design dashboard and section where developers can produce their own special workflows to decouple experimentation analysis from everyday flag monitoring and maintenance. Formal experiments allow developers to define hypotheses, duration, targeted rules and treatments for analysis.

“For example, a bank might say, we’re going to look at auto loans as an offer we’re going to show you, and the key metric we’re trying to optimize is loan originations,” Stuart said. “Are we increasing loans for the bank? Or are we acquiring new customers? So, you can see how they start to enable these new capabilities and run those experiments.”

This end-user business logic example is one of many use cases that features could affect. Code changes and application features could change processing times, cloud costs, end-user experience and myriad other metrics once released that are otherwise difficult to see in testing.

Gradually releasing them means they can be monitored safely and have a lower impact than being released all at once and therefore breaking all at once. That gives development teams the time to get ahead of potential problems before they become devastating problems.

Feature management and experiments also make running beta mode feature updates simple because the beta user base is simply a segregated group of users. Those users know that their app might not run correctly – all for the chance to play with a feature early — unlike ordinary users, who want the best possible user experience.

Image: Pixabay