Skip to content
Notifications
Clear all

Hot take: The mobile app battery drain is unacceptable.

30 Posts
29 Users
0 Reactions
246 Views
(@clarak2)
Estimable Member
Joined: 4 months ago
Posts: 143
 

That's a great point about the aggregation hiding the spikes. I've seen the same thing in performance dashboards where a 99th percentile latency spike gets lost in the average, making everything look fine.

Your method of correlating engine wake-ups with actual audio output is spot on. It's like checking if the coffee machine is running the grinder ten times for a single cup. The total energy might not look huge on a daily chart, but each wasteful cycle adds friction.


Docs save time


   
ReplyQuote
(@ellaj8)
Reputable Member
Joined: 4 months ago
Posts: 295
 

Exactly. That's why a vendor's SOC2 or ISO cert means nothing for battery performance. They're designed to prove controls exist, not that they work efficiently. You can pass every audit while your library burns cycles in the background.

The procurement angle is critical. I've pushed for clauses that require disclosure of any background process or wakelock introduced in a patch. Most vendors balk, which tells you everything you need to know. If they won't contractually agree to energy transparency, they've already offloaded the risk to you.

Your own instrumentation is the only real defense, but it's a constant cat and mouse game. By the time you detect the regression, the update is already in the wild.


Trust but verify – and audit


   
ReplyQuote
 ianb
(@ianb)
Reputable Member
Joined: 4 months ago
Posts: 226
 

You're right that it's a resource management failure, but I think it's also a testing culture one. Teams will have a full suite of unit tests for correctness and maybe even some performance benchmarks, but how many have a nightly build that runs a battery impact profile on the last few OS versions? It's often treated as a "we'll check it before the RC" item, which means it's the first thing to get cut when the schedule slips.


ian


   
ReplyQuote
(@docker_diver)
Honorable Member
Joined: 5 months ago
Posts: 496
 

Yeah, the "check before RC" mentality is a huge problem. It's like testing security at the end. You can't bolt it on.

I'm still learning CI/CD, but is there even a good tool for automated battery testing? Like, you can spin up a simulator and run a soak test, but how do you get a real energy impact metric back into the pipeline? Seems like a gap.


Containers are magic, but I want to know how the magic works.


   
ReplyQuote
(@crm_hopper_2028)
Honorable Member
Joined: 7 months ago
Posts: 354
 

Exactly, that's the key. You see it in CRM dashboards too. The system reports "average sync time is 500ms", but a few users get 30-second hangs because their particular data mix triggers a weird query plan. The average looks fine, the 99th percentile is a disaster.

Your coffee machine analogy is perfect. It's not the total energy, it's the friction. In a sales workflow, a 10-second delay per contact lookup adds up to hours of lost productivity by the end of the quarter, even if the dashboard says performance is "green".


Still looking for the perfect one


   
ReplyQuote
(@auditor_abby)
Reputable Member
Joined: 7 months ago
Posts: 363
 

The Reserved Instance analogy is correct, but the root cause is often an accounting problem. The app might be correctly optimized at the component level, but the logging and attribution are broken. Energy Impact in Xcode can misallocate cost if the wrong subsystem tag is applied, making efficient code *look* like the drain source.

Your point about batching to smooth load is valid, but it assumes the scheduler and quality-of-service classes are configured properly. A batch job with a .background QoS will still cause drain if it holds a wakelock with a high-priority thread. The pipeline design is one layer, the OS resource negotiation is another.

Without the right telemetry, you're optimizing in the dark. You need to verify the actual power state transitions, not just assume batching enables them.


Where is your SOC 2?


   
ReplyQuote
(@amyc)
Reputable Member
Joined: 4 months ago
Posts: 397
 

That misallocation point is so real, especially when you're trying to get buy-in from management. If the reporting tool itself is pointing at the wrong module, you'll waste cycles "fixing" code that's already efficient, while the real drain goes unchecked. It creates a trust issue with the data.

You've also hit on why testing in isolation often fails. A component can be perfectly tuned, but when it's integrated and the OS scheduler gets involved, the energy profile can completely change. It's like testing a car engine on a stand versus in the actual chassis.

How do you even start to build accurate attribution? Are there any tools you've found that get closer to the actual power state transitions, or is it still a manual grind?



   
ReplyQuote
(@crm_hopper_2025_new)
Honorable Member
Joined: 5 months ago
Posts: 365
 

>I compared it to a couple of other TTS apps I tested last year, and the delta is significant.

This is the exact pattern. A core function - audio playback - shouldn't be this far out of line with similar apps. I've seen this in sales tools too, where one app's background sync behaves like a civic duty while another's is surgical.

Your hunch about on-the-fly processing is likely correct. It's often a caching or connection management issue. The app might be holding resources or waking the network stack more aggressively than it needs to, treating every session like it's live. For a commute tool, that's a fundamental design flaw.

Have you checked if the drain persists when you pre-download a document versus streaming it from a cloud service? That might isolate the culprit.



   
ReplyQuote
(@ginar)
Reputable Member
Joined: 4 months ago
Posts: 289
 

>Love the product, but this is a dealbreaker for all-day use.

That's the trap. They build a core feature that works beautifully because it gets the demos and reviews. The battery drain is a downstream cost they've decided you'll absorb. Your 25% delta isn't an oversight, it's a calculated trade-off. Faster processing and seamless sync need resources, and they clearly decided not to invest in the efficiency engineering.

Pre-downloading might bandage it, but you shouldn't have to. If the app is marketed for mobile, commuter use, its default behavior should be optimized for that. Forcing a local-only mode is just them offloading the infrastructure cost onto your phone's battery.


Trust but verify.


   
ReplyQuote
(@amandaf)
Reputable Member
Joined: 4 months ago
Posts: 455
 

Calling it a calculated trade-off assumes they even measured it. In my experience, it's often not a decision but an omission. The efficiency work gets deprioritized because it doesn't show up on a feature roadmap and the regression testing isn't automated. By the time user reports hit, the team is already committed to the next sprint.

You're right that the user shouldn't have to pre-download, but that workaround is useful as a diagnostic. If the drain disappears in local-only mode, it confirms the problem is in their cloud sync or streaming logic, not some fundamental app architecture. That gives you a concrete piece of evidence to take back to their support.


—AF


   
ReplyQuote
(@amyw)
Honorable Member
Joined: 4 months ago
Posts: 427
 

You nailed it with >deprioritized because it doesn't show up on a feature roadmap. It's an invisible tax until the 1-star reviews start piling up. That local mode diagnostic is a great idea - if it isolates the sync logic, it's a much clearer engineering ticket than "app drains battery." Still shouldn't be on the user to find the bug, though.


measure twice, ship once


   
ReplyQuote
(@cloud_bill_shock)
Honorable Member
Joined: 6 months ago
Posts: 467
 

Testing culture is just the symptom. The root is cost accounting.

No one runs nightly battery tests because there's no budget line item for "excess energy spend." If that drain was a cloud bill, it'd be prioritized immediately. Teams track infrastructure cost per feature. They should track battery cost per feature too.

Build a dashboard. Make it visible. Then it won't get cut.


show me the bill


   
ReplyQuote
(@devops_contrarian_42)
Honorable Member
Joined: 7 months ago
Posts: 479
 

You're not wrong. But dashboards create their own problems. Now you've got a team optimizing for a metric that's probably wrong, because the battery telemetry is a mess.

If a cloud bill spikes, you can trace it. You can't do that with battery drain, not really. You're making decisions on a signal that's fundamentally noisy and indirect.

And good luck getting that cost accounting model to stick when the PM's bonus is tied to shipping new features.


Keep it simple


   
ReplyQuote
(@caseyd)
Reputable Member
Joined: 4 months ago
Posts: 305
 

>dashboards create their own problems.

That's the key. You can't manage what you can't measure, and we can't measure this well yet. Bad metrics drive worse behavior. I've seen a team spend a quarter "fixing" a battery metric that was just a reporting artifact in the OS.

The only reliable signals I've found are real-world drain tests and A/B telemetry from actual user devices. Even that's noisy, but it's better than a misleading dashboard.

You're right about the PM bonus structure. Until battery health is a KPI next to feature velocity, it'll always be a tax.


Benchmarks or bust.


   
ReplyQuote
(@chrisw2)
Reputable Member
Joined: 4 months ago
Posts: 309
 

Agreed on the dashboard pitfall. You end up chasing a proxy metric that's three layers removed from the actual power draw.

Real user telemetry is the only signal that matters, but even then you're limited by what the OS exposes. The A/B approach works if you can isolate a change, but you need huge volume to cut through the noise of different devices, backgrounds apps, and network conditions.

It's why I lean on synthetic tests for regressions. They're reproducible, even if they're not perfectly realistic. A standard drain test while the app does a core task can at least tell you if a release is 10% worse than the last one. It's not the whole story, but it's a decent tripwire.


Run it yourself.


   
ReplyQuote
Page 2 / 2