| p>It is odd how errors are participated form | | | | values |
| perspective of end data user and from IT side. From | | | | What? |
| IT side it is normal to see mistake and to fix it but from | | | | Ever had situation where in reporting software were |
| the users perspective it is always a big riot effect. End | | | | several errors, not one. And errors might accumulate |
| users do not know nature of errors, do not see | | | | their difference value or could have positive and |
| everyday problems in maintenance process, and | | | | negative error values - meaning, compensating |
| especially do not know what effort from IT side is | | | | difference and misleading reporting error analysis. It is |
| involved for error fixing. End users do see figures and | | | | more often situation then thought. |
| they see good or bad figures. End users also have | | | | From perspective of business users when seeing |
| obligations to present data to management. Those | | | | sudden decreases and raises in data they would on |
| who have to answer unpleasant questions about | | | | the first ball say that |
| figures, are far more nervous then IT who is not | | | | - "IT is once again playing with data" |
| involved in reporting stress. | | | | - "IT is not acting responsible about consequences" |
| What causes errors in Business Intelligence systems? | | | | But truth is more often it is very hard to fix errors in |
| - Most common error source is code structure | | | | live production systems, system has to run and needs |
| error. Code does not support new data structure, | | | | cardiovascular operation. So it is usually overnight |
| new functionalities. | | | | working time. Furthermore, there is no adequate |
| - Manual data input. Personnel can entry free style | | | | resources to make parallel work and to test |
| description... | | | | everything in proper environment. And at the end IT |
| - No validation at point of entry. If something is wrongly | | | | fixes errors one by one over time and this time |
| inserted there is no validation mechanism to alert. | | | | period is not only one night or one day or wished time |
| - Implementation error. Happens from time to time... | | | | frame from business side. So, if there are several |
| - System error. Upload breaks at half and figures are | | | | "compensating" errors during data cleansing time... off |
| presented in reports... | | | | course business side should expect odd play of |
| Are errors truly errors or something else? Mostly they | | | | figures. |
| are result of natural business growth that is not | | | | Try to explain that within official reporting toward |
| supported within applications. | | | | stock and try to revise official published figures |
| So, don't worry if they happen. Worry about resources | | | | backwards. |
| for permanent maintenance and upgrade. | | | | Got the point? |
| Complementary errors - several errors compensating | | | | |