Official fuel consumption and range figures are consistently better than what owners achieve. The gap is not primarily deception, but a consequence of what a standardised test is designed to do.
What the test is for
The purpose of a certification test is comparability. Every car must be measured identically so that figures from different manufacturers mean the same thing.
That requires eliminating variables, so testing is done on a rolling road at controlled temperature, following a defined speed profile.
Comparability and realism pull against each other, because real driving is precisely the set of variables the test removes.
Why the conditions flatter the result
Test cycles use gentle acceleration and steady speeds, avoiding the hard acceleration and braking of ordinary traffic.
Ambient temperature is controlled, so the engine reaches operating temperature quickly and no energy is spent on heating or cooling the cabin.
Wind, gradients, road surface texture and rain are all absent, and each of them costs energy in the real world.
What the driver contributes
Consumption is highly sensitive to acceleration and to speed, since aerodynamic drag rises with the square of velocity and the power needed rises faster still.
Short journeys are especially costly because a cold engine runs richer and its friction is higher until it warms.
Two drivers in identical cars over the same route can therefore produce results differing by a wide margin without either doing anything unusual.
How the tests were tightened
Earlier procedures used low speeds and very gradual acceleration, and permitted preparation that had little to do with a showroom car.
Newer procedures raised speeds, extended duration, and account for optional equipment that affects weight and drag, which brought figures closer to observed use.
On-road testing with portable measurement equipment was added specifically because a laboratory cycle can be optimised for in ways real roads cannot.
Why the figure is still worth reading
As a prediction of personal consumption it is unreliable, and treating it as one leads to disappointment.
As a comparison between two cars measured under identical conditions it remains valid, because whatever flatters one flatters the other equally.
The useful approach is to treat the number as an index rather than a forecast, and to apply a consistent personal correction based on what previous cars actually returned.
That correction is reasonably stable for a given driver and set of roads, which makes it a better predictor of a new car's consumption than the certified figure on its own.