MarketPulse: How it Works

MarketPulse applies pattern recognition techniques used in physics and engineering to infer market response curves that express the impact of specific marketing actions in the past (such as advertising spend, distribution changes, or price changes).

The process works as follows:

Your sales data (for example, units sold or monetary value) for a weekly or monthly period over the past are fed into MarketPulse
Time series for the same period, for any number of marketing activities, are also fed into MarketPulse, using any units available (such as spend, share of voice, or TVRs)
Through a combination of automatic optimization (genetic algorithms) and human guidance/intervention, MarketPulse is used to deduce the underlying response curves of the various marketing actions in order to create a statistically satisfactory set of profiles
Gaps between the predicted and actual results are analysed to determine additional factors that need to be considered (such as competitor actions or seasonal effects), resulting in a more robust model
The resulting model is used to plan future marketing plans
Over time, the process can be repeated to yield a richer set of profiles and more accurate and effective plans

Our proprietary, web-enabled software performs these operations in real time, can incorporate multiple data feeds (for example, your databases or spreadsheets), and allows you to view results at your own pace or requirements.