When choosing a new smartphone, there are several factors which can influence your decision including the price, the camera, the brand, and the overall system performance. When measuring performance, there are two popular methods used: speed tests and benchmarks. The latter is an app that runs on your phone and performs various complex computations and then produces a score. The problem with these “complex computations” and the final “score” is that they can leave users feeling that the test is too clinical and removed from real-world usage.

At the other end, speed tests try to simulate real-life usage by starting a series of popular apps, one after the other, and seeing which device is able to get through the series fastest. The problem with speed tests is that app startup times are not a good indicator of overall app performance. Consumers need to know how well an app actually performs, not how quickly it starts. A complex 3D game might start quicker on device A, but device B actually offers higher frame rates and more graphical details.

App startup times are not a good indicator of overall app performance.

To bridge the gap between these two extremes, I have devised a new system called “Speed Test G.” I have written 10 standalone Android apps which can be run independently of each other. Each app performs a task or simulates a real-world usage scenario and then exits. I have also written a replacement launcher which has but one function: to start these 10 apps, one at a time, and measure how long the whole test run takes.

The result is a system which combines the best parts of speed tests and benchmarks. Because the apps are independent, they need to be loaded into memory and initialized just like any other app, just like a speed test. The apps then perform CPU- and/or GPU-intensive tasks, similar to the way that benchmarks work.

At the end, the launcher presents the overall test run time. This isn’t a weighted score, nor is it relative to some baseline — it is a measure of the time taken. Simple, reliable, and easy to compare.

For more details on how the tests work and for a longer discussion of the benefits, be sure to watch the video.

Watch: 5 reasons why smartphone speed tests are fundamentally flawed