3.2. Controlled or open process?

As the DHIS2 customisation process often is and should be a collaborative process, it is also important to have in mind which parts of the database are more critical than others, e.g. to avoid an untrained user to corrupt the data. Typically it is a lot more critical to customise a database which already has data values, than working with meta data on an “empty” database. Although it might seem strange, much customisation takes place after the first data collection or import has started, e.g. when adding new validation rules, indicators or report layouts. The most critical mistake that can be made is to modify the meta data that directly describes the data values, and these as we have seen above, are the data elements and the organisation units. When modifying these definitions it is important to think about how the change will affect the meaning of the data values already in the system (collected using the old definitions). It is recommended to limit who can edit these core meta data through the user role management, to restrict the access to a core customisation team.

Other parts of the system that are not directly coupled to the data values are a lot less critical to play around with, and here, at least in the early phases, one should encourage the users to try out new things in order to create learning. This goes for groups, validation rules, indicator formulas, charts, and reports. All these can easily be deleted or modified later without affecting the underlying data values, and therefore are not critical elements in the customisation process.

Of course, later in the customisation process when going into a production phase, one should be even more careful in allowing access to edit the various meta data, as any change, also to the less critical meta data, might affect how data is aggregated together or presented in a report (although the underlying raw data is still safe and correct).