Performance of other wholesalers
A pharmaceutical wholesaler wanted to find out more about its market performance compared with the performance of other wholesalers in the industry. The data required to calculate the market shares were only partially available in the wholesaler’s source system, namely the order files.
Negotiation took place at board level
In order to achieve exchange of sales data – anonymous of course – between wholesalers in the same line of business a negotiation took place at board level. As a result, the wholesalers established a joint foundation with the aim to once a month process all sales data – according to a specific format – into one large file containing all revenue data divided by product level and month level.
Linking the data at product level
It goes without saying that the customer level – for competitive reasons – was removed. The wholesalers would then load the file into their data warehouses in order to create a coupling – at product level – between their own revenue data and the revenue data of the other wholesalers. In this way, they could trace and report their market share on a monthly basis.
Paid or unpaid
In many other industries similar foundations have been set up in order to allow organizations to use – paid or unpaid – revenue or customer characteristics of a certain industry or region, mostly aggregated and anonymised. In the context of possible sources for external (demographic) data, we should also think of data suppliers such as Wegener DM marketing, TNT, AC Nielsen, Statistics, the Chamber of Commerce, Cendris, Acxiom etc.
Integrating external data is more difficult
Coupling external data turns out to be much more difficult than integrating internal data. This is mainly because the data definitions also need have to be aligned beyond organizational boundaries.
The website of a competitor
It becomes even more complicated when the external sources are rarely structured, if at all. This may be a competitor’s website from which we can distil, for example, that a new product has been launched. In this context, we will need to add a so-called text-mining model to our data warehouse. This model (almost) immediately converts unstructured data into information and knowledge if setup properly.