TABLE OF CONTENTS Experiments are used when some decision is under consideration and the results cannot be predicted from the existing experience either of the organization itself, or of the activity of other organizations within the marketplace. Experiments attempt to measure change under, as far as possible, normal marketplace conditions. The problem is that they are expensive and can disclose plans to competitors. It may also be difficult to achieve normal marketplace conditions on a sufficiently small scale for the exercise to be considered an experiment rather than a suck-it-and-see trial. In addition, retailers are often unwilling to co-operate with a test in one area alone. Experiments should normally be confined to situations where the expense and effort involved are calculated to be more than compensated for by the outcome of the experiment. This should be estimated beforehand. It should also be determined that the information could not be gained any other way; for example, by monitoring similar approaches by other companies, comparison with other markets, or reanalysing existing data from past or syndicated surveys. Given that experiments are thought to be worthwhile, several approaches are possible.
Two types of market experiment can be used before new product launching: experimental launching or pilot launching. For existing products, market tests may be either specific or exploratory. If an experiment is to be carried out, then the type of experimental design to be used must be decided, and this may be ex post facto, split-run, before-and-after with...
