How Canal+ cut app startup time by 40% on low-memory devices with the Kotzilla Platform

TL;DR

• Slow start-up on low-memory devices was hurting subscriber retention in Canal+'s growth markets
• Kotzilla analyzed 120M+ sessions and helped cut start-up time by 40%
• Kotzilla is part of the Canal+ release process: every version baselined, AI-assisted fixes generated from Kotzilla's runtime context

The app plays a central role in delivering fast and reliable access to payment services, where performance is critical for both users and support teams.

1. The challenge

Canal+ is a global media and entertainment company with 26.8 million subscribers and over 400 million monthly active users across its streaming platforms. The group is growing fast in emerging markets where budget Android devices are the norm.

• Performance on budget devices is a business problem: on low-to-mid-range hardware, a slow or unresponsive app directly threatens subscriber retention and wastes acquisition spend.
• Complex multi-variant ecosystem: 17 app variants across mobile, TV and partner set-top boxes require separate monitoring under shared package names.
• Traditional tools fall short: existing APMs show that startup is slow, but not why. Root cause analysis requires manually cross-referencing crash reporters, CPU profilers, and logging tools.

2. The Kotzilla solution

The Kotzilla Platform is built on Koin, the #1 dependency injection framework for Android (40% of native apps). This gives the platform something no other observability tool has: direct access to the app's architecture at runtime. Where other tools report symptoms ("startup is slow"), Kotzilla identifies the root cause and generates fix guidance so developers can resolve issues in minutes instead of days.

For Canal+, this translated into:

• 40% faster startup (2+ seconds) on low-memory devices, even after Android Baseline Profiles: root causes identified and fixed within the same development sprint
• 120M+ sessions captured across 17 app variants and 2,300+ release versions, embedded in the standard release process
• AI-assisted fixing is part of the workflow: the team runs Kotzilla-generated prompts, loaded with the app's real runtime context, in their coding assistant, and the resulting fixes ship to production.

3. Business impact

Canal+ invests heavily in subscriber acquisition through telco partnerships and device bundles. When the app performs poorly on the device a new subscriber just received, that acquisition cost is wasted. In markets where global and local streaming players compete on the same hardware, the app that loads fastest wins.

And the gains are not a one-off: every release is baselined against the previous one, so regressions are caught before users notice, and adoption is extending beyond the core dev team to QA and CI/CD workflows

Find out why your app starts slow, and on which devices. Then fix it with the Kotzilla MCP Server.