Delivery Hero is standardizing on Sentry for mobile error and performance monitoring, giving engineering teams autonomy, accelerating performance improvements, and reducing noise for platform teams.
Introduction
Delivery Hero is a local delivery platform that works with a global network of restaurants, local shops, partners, and riders in approximately 70 countries. With a company mission to deliver anything to customers fast, providing users with a great mobile experience is a business-critical function.
The platform engineering team at Delivery Hero is responsible for providing support and tooling to the app teams doing native mobile development on iOS and Android apps. Specifically, they support Delivery Hero’s foodora, foodpanda, and Yemeksepeti brands, which serve markets across Europe and Asia. That means ensuring the engineering teams have the right tools to support reliability, mobile releases, and performance across multiple apps serving different markets.
Challenge: Fragmented alerting and inefficient triaging
The tool previously used by the platform engineering team for crash reporting wasn’t able to scale to fit the needs of an organization of Delivery Hero’s size. The result was an inefficient debugging process:
- Lack of issue grouping: The tool captured crashes, but couldn’t effectively group issues together, leading to wasted time manually triaging related errors.
- Multiple unconnected projects: Because of the way the engineering org is set up at Delivery Hero, they had to set up multiple unconnected projects and swivel-seat between them to get visibility across the engineering organization.
- Basic alerting: Alerts lacked customization and filterability, leading to noise and creating issues when trying to scale.
- Errors unconnected to owners: The team lacked visibility into who was responsible for fixing errors, leading to wasted time tracking down who to route issues to.
Because they owned the tool, the platform engineering team had to manually manage the triaging of issues as they came in, taking time away from their core responsibilities and delaying how long it took for the application teams to deliver fixes:
*“It’s hard to know who’s on call for each team, which channel to use, and who to contact. Asking one team to manage all of that adds extra work and slows things down, when the issue could have gone straight to the owner.” - *Pero Smiljkov, Senior Engineering Manager, Platform Engineering @ Delivery Hero
As the number of organizations that they supported grew, the platform engineering team needed a more efficient approach to monitoring at scale.
Why Delivery Hero chose Sentry: Enterprise-level scalability and efficiency across engineering teams
Sentry helped the platform engineering team take themselves out of the issue alert cycle:
-
Granular control over alerting and issue grouping flags only relevant issues and sends them to the right teams, reducing noise.
-
Using Suspect Commits, the team is able to define custom ownership rules to route alerts directly to the teams responsible for the fix, decreasing the time required to triage and address issues.
-
Getting fast alerts on issues like ANRs helps the application react quickly and resolve issues on core screens like cart and checkout (where issues can contribute to order loss)
“Sentry gave more flexibility, like narrowing down a problem to one country, subgroup of the users, or to an experiment or feature flag that we’re running. It became much easier to search, filter and navigate through the errors.”
— Pero Smiljkov, Senior Engineering Manager
-
Using Release Health, the team is able to get visibility into crash-free session rate and other critical metrics with a single-pane-of-glass release dashboards across multiple brands/apps
“The release dashboard is quite useful. We are using it to track new releases and then I can see the crash free users and metrics for all the different apps that we have. That’s helpful to my team because we’re owning the mobile releases and the overall quality of the app.”
— Pero Smiljkov, Senior Engineering Manager
Tracking performance across critical app experiences
Beyond increasing the efficiency of error monitoring, Delivery Hero has used Sentry to help improve the performance of their mobile apps across the brands supported by the platform engineering team. With Sentry, they can track a suite of metrics like Time To Interactive (TTI) across the core screens of Delivery Hero apps while in Beta. This helps the app teams improve the performance of critical app experiences when releasing new versions of the app.
The impact
Using Sentry, Delivery Hero has seen:
- Faster time to detection (MTTD): MTTD has gone from several days to 2 hours for slow-burning errors, and direct alerting to app teams avoids delays from platform team triage
- **Improved crash-free session rate: **Monthly crash-free user sessions have gone from 99.77% to 99.95% on Android and 99.82% to 99.95% on iOS.
- Improved triaging: GitHub & Slack integrations made debugging faster for application teams
- Better quality gates: Beta performance regressions are caught before release using performance monitoring
- Visibility across mobile releases: The team can track critical metrics and regressions across app releases (around 50 releases per year per platform) from multiple teams in one place.
Sentry is now the default mobile observability tool across Delivery Hero, and the team plans to expand Sentry’s performance monitoring capabilities into production, enhancing anomaly detection across business-critical flows.