It is fairly clear that many of the arguments against proposition testing are really arguments against propositions themselves.

A campaign, like a brand, is not just a number of bits put together-a claim here, a pack shot there, a reason why somewhere else. If we try to produce it by the atomistic approach, we will end up with a sort of Identikit brand. It will be a perfect description of the structure of the brand, as the Identikit can describe the contours of the face. But it won't be the same thing. The brand will never come to life.

The only leverage the manufacturer can apply to the retailer is his relationship with the consumer. And the main element in profit growth is going to have to lie in making his brand more valuable to the retailer, through its being more valuable to the consumer. And that means his brand must be unique, it must have no adequate direct substitutes - because it is in this, after all, that value lies.

Advertising tends to be most effective in jogging finally into action those people who are well-enough disposed towards a product, but have not yet got around to buying it.

There are really very few economies of scale in centralising or standardising advertising.

Research can only present data about the past. No one seriously believes that people's answers to hypothetical questions about the future accurately represent their future behaviour; they merely represent a current attitude, which may or may not be translated into future behaviour.

One can often trace the sources of a brand personality-here it is the advertising, there the pack, somewhere else some physical element of the product. Of course, the personality is clearest and strongest when all the elements are consistent.

Tracking account planning is rather like counting a mixed batch of tropical fish. You think you see patterns, but they've all changed by the time you've finished counting.