The tests are run without air-con, lighting or audio equipment and, because the tests have to accommodate commercial vehicles, the acceleration components allow 50 seconds to reach 62mph, which is not how most people drive.
All of which I’m sure is very scientific and repeatable, but is it representative? We certainly very rarely get close to the claimed figures in our tests, which may not have the total accuracy of a lab test, but I think are more thorough and realistic of what you, as an owner are likely to achieve.
In our full road tests we currently report three fuel figures, each of which is calculated by measuring the actual fuel put into the tank (we also record the trip computer values for comparison). Our track figure represents the worst-case scenario, i.e if you absolutely hoon it everywhere. We calculate this at Mira, during the performance tests. The distance covered does vary from car to car, depending on, for example, the transmission type or how many attempts it takes us to get the best acceleration figures, but it always exceeds 50miles.
Our touring figure is recorded over a fixed route, including accelerating gently (but not so gently that you’d be a danger to other traffic) to a 70mph cruise, but also with a shorter section at 50mph.
Our Average figure is total fuel economy while we have the car, typically over a week, and covering in excess of 500 miles.
In addition we’re thinking of adding an extra figure, recorded at Mira, for the car travelling at a constant 80mph. Not that we’re recommending you break the law.
With the increasing focus on diminishing resources and environmental impact, perhaps it's time the official tests came in line with reality. In the meantime you know where to come for the real story.