arrowCustomer Stories

Fortune 100 Health Insurer minimizes downtime for members with controlled, automated software releases

Before

Manual approval workflow that slowed urgent production fixes

Fragmented audit trails created compliance risks

Releases were commonly issued at night to avoid customer impact

Monthly release capacity was capped at 60-70

After

95% of the release process is automated

Immediately scaled to 150-200 monthly releases without additional headcount

Change failure rate below 4% with zero customer-impacting incidents

Mobile releases ship on average 2 days faster per sprint

About

This organization, part of a Fortune 100 health services provider, delivers digital healthcare experiences to more than 12 million members across web, mobile, and backend platforms. The release engineering team supports code changes and releases for dozens of microservices and feature teams across a large multi-platform product.

Their work spans backend services, mobile applications, and infrastructure platforms, ensuring secure, compliant, and reliable software delivery in a regulated healthcare environment.

Challenge

A Senior Release Manager on the release engineering team recognized that the legacy release process was a liability. The system lacked speed and safety. When something went wrong, rollback required manual intervention and, sometimes, hours of work. 

Release approvals relied on a manual, multi-step workflow involving Word templates, email chains, and copy-pasting across multiple systems. Engineers filled out forms by hand and posted them to Teams, where each approver had to respond individually. They then transferred those approvals into Rally and Quickbase, pasted screenshots into specific fields, and emailed executives for final sign-off. For audits, they often had to dig up approvals across systems and replicate the same data in multiple places, often doing these tasks ten times over for a single release. It wasn’t unusual for the approval process to take longer than the coding itself.

/ /

We realized that one person could support roughly 40 releases a month with the old process. We considered how much overhead we spent on rework or duplicated effort. All that takes time and a significant amount of overhead for really no gain.

Sr. Release Manager

Fortune 100 Health Insurer

Solution

To modernize the release process, the team began by building a new automation layer using Power Automate, ServiceNow, and SharePoint. That removed 95% of the manual workflow required for approvals and created an auditable system of record.

/ /

It's not uncommon for me to literally just go get a glass of water and come back to three or four releases that are now scheduled that I didn't know about or didn't exist before I got up from my desk. I literally just clicked the approve button, we got a change number, and we're going to production in the next two to three minutes.

Sr. Release Manager

Fortune 100 Health Insurer

The release management team uses LaunchDarkly to ship code to production at any time using the automated approval process, and then control when and how that code is exposed. They can gate features behind internal approvals or testing checkpoints, gradually roll out changes to specific user segments, and instantly disable functionality if an upstream dependency fails. This level of control enables safer testing in production and rapid response to risk, all without the need to revert code.

/ /

Some of our customers want to have all the new things, and some want to be the last group to see those things. Feature flags allow us to control who can see what features in production.

Sr. Release Manager

Fortune 100 Health Insurer

LaunchDarkly is gaining more widespread adoption across frontend, backend, mobile, and infrastructure teams. Product owners using feature flags can manage feature exposure without needing additional engineering work, and teams commonly toggle flags during daytime hours rather than deploying late at night. Flags are increasingly treated as a standard part of the release process, resulting in safer releases, fewer incidents, and considerable time savings.

Results

The release team scaled from 60-80 releases per month to 150-200 releases per month while maintaining a two-person team. Most releases happen during the workday, thanks to the assurance provided by LaunchDarkly granular controls.

Key outcomes include:

  • 160-320 engineering hours saved each month after Power Automate workflow and LaunchDarkly implementation
  • 3.91% change failure rate with very few issues impacting customer awareness because of rapid rollbacks
  • 3-5 minutes from approval to deployment, decreased from several days
  • 2–5 hours saved by each team per month on DORA reporting tasks
  • 5 - 10 rollback events avoided each year

The integration with SharePoint and Power BI enables one-click DORA metrics reporting across 50+ applications, helping teams measure and improve their delivery pipelines. Flag usage is tracked and standardized, creating a foundation for even more advanced strategies like A/B testing and automated rollbacks.

/ /

As a release manager, better automation and LaunchDarkly have freed me up to actually ask real questions. Now I can actually dig in and ask, what is the real risk for this release? I can actually know my customer a lot better because I'm not spending 90 percent of my time cutting and pasting.

Sr. Release Manager

Fortune 100 Health Insurer

This new process isn't just a technical upgrade. It's a model for delivering at scale without compromising safety, compliance, or speed. And it's replicable.

/ /

Figure out what your biggest pain point is, automate that, and see the quick win. Then you realize, wow, this just took out 35 minutes of my release time, and ask, ‘what do I spend 5 or 10 minutes on?’ Is it worth it for me to spend 40 hours figuring out how to automate that? Yes, because that time gets reclaimed very quickly.

Sr. Release Manager

Fortune 100 Health Insurer

With feature management and automation at the center of its delivery process, the team has proven that it can scale releases, reduce risk, and support more teams without adding headcount. Their vision is a future where deployments happen continuously, features are always behind flags, and releases can be flipped on safely at any time of day.

/ /

My goal is that as soon as something gets merged into GitHub, there is an automated push a pull request to production within minutes of it being done, and that everything would be behind a feature flag. We’d push to production as a matter of course, and just use LaunchDarkly to decide when the world sees it.

Sr. Release Manager

Fortune 100 Health Insurer

More case studies