I have been working on battery consumption analysis for my application, and as part of this effort, I wanted to understand how competitor apps behave under similar usage conditions.
To do this, I downloaded competitor apps from the App Store and attached them to Instruments via Xcode. I then executed a defined set of manual test scenarios to simulate real user behavior. During these tests, the iPhone was connected to a Mac and charging continuously, which meant that System Power Usage logs were not generated in Instruments.
However, I was able to capture detailed metrics such as:
Network usage
CPU load
GPU activity
Display and brightness impact
Other runtime performance characteristics
Since direct battery drain data was unavailable, I used derived analysis (with AI assistance) to estimate approximate power consumption based on the above metrics, assuming real-device (battery-powered) conditions.
According to Apple documentation, System Power Usage in Instruments is not directly tied to the device’s battery percentage. Instead, it appears to be computed using contributing factors such as CPU, network, display, and other subsystem activity. This raises a few important questions about data reliability and methodology.
Key questions:
How reliable are Instruments-based metrics (CPU, network, display, GPU) for estimating real-world battery consumption when the device is charging?
Can these metrics be safely used as a comparative baseline between competitor applications, even if absolute battery drain values are unavailable?
Is the System Power Usage instrument essentially a derived model based on subsystem activity, and if so, does it remain accurate when the device is not discharging?
From Apple’s perspective, is this a valid approach for relative power comparison, provided that:
The same device is used
OS version is identical
Test scenarios are consistent and repeatable
Based on these findings, would it be reasonable to proceed with instrumenting our own application, run the same scenarios, and draw conclusions using relative comparisons rather than absolute battery percentages?
The intent is not to claim exact battery drain numbers, but to establish a directionally correct and repeatable comparison that can guide performance optimizations in our own application.
I would like to understand whether this methodology aligns with Apple’s recommended practices, or if there are limitations or inaccuracies I should be aware of before relying on these results for decision-making.
