Accuracy depends on the testing method, testing equipment and testing technician.
When we test a battery, we want to answer the question: Will this battery provide the volts and amps required to turn a starter fast enough and for enough time to start the engine under the worst expected temperature conditions?
There are two main testing methods: digital (computer-based) testing and high amp testing.
Digital testing involves a tester that looks like a multimeter with small battery clamps and thin wires, perhaps as thick as a typical USB cable. The technician enters any required battery specifications into the device. When the test begins, the device measures voltage very accurately and places a small load on the battery, and measures voltage change with the load. Then the internal computer consults its algorithm, refers to its map of all possible voltage curves, and reports good, weak or bad. It may also report the cold cranking amps (CCA) that the battery should be able to produce.
This method is not perfect. It is not an actual real-world test of performance, but a projection or prediction based on a computer model that usually works, but not always. There are edge cases that clearly fall outside the capabilities of a computer model to predict performance. We have had such cases on this Stack where digital testing produced false "good battery" results when the battery was clearly bad.
The other testing method, the high amp method, is 100% reliable because it is not a simulation or projection, but an actual, severe test using the same or more amps that are required to start a car under difficult circumstances.
A high amp tester sometimes comes on wheels, or it can be portable with a handle like a small briefcase. It has large battery clamps with high-force springs that might take 2 hands to open. The battery cables are thick like jumper cables. For the test to be valid, the battery must be fully charged, and this may be the reason that high amp testing is less common today: if the customer has to wait 1-2 hours to get results while the battery fully charges, it's not great for business.
The test is performed by adjusting a knob or crank until the tester is drawing 50% of the battery's CCA. So for example, if it is an 800 CCA battery, the tester is drawing 400 amps. This high load is maintained for 15 seconds, and then the voltage is measured at the end of this 15 seconds just before the load is disconnected. That voltage is compared to a table of voltages at various temperatures, and the results are reported as good, weak or bad. (Some models of tester may use slightly different amps or time standards, but the principle is the same.)
So you see that this test shows actual performance of the battery as if it is cranking a cold engine for a long time. A clearly bad battery will have its voltage fall below 10 volts as soon as the heavy load is started. An old battery nearing the end of its useful life will have its voltage drop slowly toward 10 or 10.5 volts as the 15 second load progresses. A new, good battery will maintain a steady high voltage for the duration of the test.
If you want to be 100% certain that you have a good battery, start with a fully charged battery and insist on a high amp test.
