I've found the Apple laptops to be closer to the actual stated times more than any other manufacturer. But I guess this is more to do with the test suites available (none for Mac). Some manufacturers are notorious (hmmph Acer) for gaming results by reducing screen brightness to 20%, running the proc at the lowest level, disabling Wifi and not much else. Basically letting the laptop sit there until the battery runs out.
I called technical support about this because I was getting ~45 minutes even though the box stated 2.5 hours. What I was doing (web browsing over wi-fi, in the late afternoon) would get me the reported time. Seriously, the only way I was going to get 2 hours would be by sitting in the dark with no internet connection using notepad.
At the end of the day, you need a 3rd party to perform benchmark tests and publish the data. It's not necessary for the manufacturers to do it, as long as it's available. The 3rd party in question could sell their data to CNET, etc., as an income stream.
An industry standard test that each manufacturer runs themselves is too hard for a computer. Too many variables. For example, the FTC has a test for speaker manufacturers to measure speaker power. What do I define as full power? I can make the speakers go really, really loud, but they sound terrible. Where do I take my power measurement? The FTC defined 10% THD+N as the max distortion allowed, so your power measurement was how much RMS power you could produce at 10% THD. However, what happens if you have a subwoofer and satellites? How do you get them all to max power at the same time, since for high frequencies, the woofer isn't doing anything, and for low frequencies, the satellites aren't doing anything. Well, you mess around until you find that magic frequency at your cross-over point where both sats and sub are at 10% THD. OK, fine, but when do you measure the power? As things heat up, amplifiers and wire and speaker coils get higher resistance, which lowers your output power. Somebody wanting to game the system could take their measurements quickly, while everything was cold. OK fine, the FTC set forth that the system had to have been running for 10 minutes at full power, before measurements are taken.
So, this is just for a lowly speaker system, and we've had to go through all of this. This is the test that the manufacturers run themselves and advertise their product with. Even with all this, there are companies that don't play by the rules. They give instantaneous power, peak power, or some other crazy power that they dreamed up over sushi and sake. Many don't bother to even list a power rating of any sort. Bose comes to mind, but there are others. Apple didn't do this either, perhaps they do now.
So, now for computers, imagine how difficult it would be, when you have variable wifi power, transmission bandwidth, distance from router, router power, variable monitor brightness, display resolution, refresh rates, variable hard drive access profiles, variable CPU loading profiles, etc. There's no way you can trust the manufacturers to not game a system so complex.
A 3rd party IS doing it and they do distribute their reports. Consumer Reports has been doing testing like this on a variety of stuff for over 70 years. They don't take outside advertising or free samples to avoid bias. Their tests have held up in court as valid. They're really good.
Next time you see a sketchy battery life claim by Apple [1], Acer [2], or anyone else, just take a glance to see if Consumer Reports has said anything about it. I'm not asking for you to take their word for granted, just consider it the opinion of a reputable 3rd party.
Yep, they do it. But they don't, apparently do it enough to please AMD.
Having somebody decide to test some random product when the feeling strikes them, is different from somebody who is religiously testing all available products creating a database where all the data fields are the same, all the test conditions are the same, etc.
From a Electrical Engineering standpoint, any battery of tests will not hold up due to the amount of variables, both software and hardware. The article references iPhone as a perfect case, but Apple (and AT&T, indirectly) have 100% control on what occurs on the device. The software, hardware, and user flow is under strict standards of processing and memory allocation. Furthermore, the low processing power and subsequent need makes the margin-of-error and standard deviation lower.
This is not the case for laptops. Due to running different programs, there are differing memory and processing allocations which cause different hardware elements to be accessed. Chrome and Firefox (same use case) have vastly different memory footprints and certain versions of Photoshop (same program) will utilize the GPU, increasing current drain. Since programs are constantly under development and will use different amounts of resources, a manufacturer cannot say that browsing the internet or using a specific program will be a certain amount of time.
Mainstream laptop batteries are Lithium-Ion, which degrades with time (faster with use). Unless retailers begin to throw out older batteries (by manufacturing date), there will be no standardization for consumers.
Most PC manufacturers might use MobileMark, but Apple tests with 50% brightness, Wifi on browsing "various websites" and editing text in a word processor. Even other than the iPhone, Apple seems to do better than other manufacturers at battery testing and disclosure.
I called technical support about this because I was getting ~45 minutes even though the box stated 2.5 hours. What I was doing (web browsing over wi-fi, in the late afternoon) would get me the reported time. Seriously, the only way I was going to get 2 hours would be by sitting in the dark with no internet connection using notepad.