Production data errors in reports

Getting nonconsolidated data with errors from production systems

Fixing errors

Occurrence of production error data for reports is very often since production systems are more process oriented then process-reporting oriented. Productions systems are customized for processes and for plain and simple reports mostly for users in divisions that are owners of system. What is usually forgotten and perhaps is not so obvious is that very aggregated reports are not included into debate during project analysis and later in deployment and implementation. Result is that aggregated KPIs from one source or from several sources are not “consolidated”. This is a financial term but fits in this perfectly. KPI A combined with KPI B and KPI C should give logical result, zero. For example if X number of subscribers deactivate subscription, Y number of subscribers purchase subscription, Z number of subscribers are deactivated do to many reasons (nonpayment, technical reasons and etc.), total number of subscribers between two months should correspond to relation of X, Y and Z. But this is not the case…

There could be many reasons:

• customers can not be kept for hours on points of sale in order to gather very detailed information to feed our production systems
• validation rules are not propagated till selling persons and sales can enter mistakes into systems
• other systems that are connected to data collecting systems generate all together data loss
• and so on …  and so on

On the top management is frustrated without having proper data gathering.

Fixing data production errorsTip: Think of taking snapshots of data feeds, with off course adequate set of attributes and additional data,  for two reporting periods and then analyze data changes and what is really happening in systems. Those changes that are not taken through validation process are causing troubles. even within millions of raw data lines can be found relatively low amount of data changes between two reporting periods. Buy knowing what happened and what caused this change implementation of explanation into reports is only matter of time.Tip: Think of storing snapshots into DWH, data mart or similar…