Configuration and Customisation

Previous

Carry out Trial Conversion

Next
Progressively refine the conversion of data from the old to the new system
  • Rectified data in the existing system
    1. Take a snapshot of the existing system data.
    2. Convert the data.
    3. Investigate data that did not convert.
    4. Investigate the new data for errors.
  • Data ready for a final conversion

There may be more than one trial conversion depending on the complexity of the conversion. An initial conversion may also be undertaken in order to identify data errors or modifications required. Errors fall into a number of categories:

  • Data which failed to convert.
  • Data which converted but is in a form that is not usable by the new application.
  • Data which converted but is wrong.
  • Data which is missing from the conversion and was not part of the conversion plan.

Data which failed to convert. This data will typically be identified in error logs or reconciliation between the snapshot of the old system and the new. In order to do the comparison, various reports may need to be written or run. For example a report on the number of product records in both systems may show one or more missing. This will require further investigation.

Data which converted but is in a form that is not usable by the new application. Sometimes this is because the data was truncated. The old field allows 100 characters and the new only 50. Very often the only way to do this is to do a straight comparison. If you do know of particular fields that stretch the limits in the old system, it is worth focusing on these in the new system. A product that has a description that is twice the normal length would be a good one to see if it converts.

Another type of error may relate to numbers or dates. The format used in the old system may be different to the new. For example (and this is a real situation), the old system may store dates as 2009-01-01 in a text field. The new system may store it as a date field and make some strange assumptions.

Data which is converted but is wrong. This category includes data converted into the wrong field. It might be a long description in the old field converted into a short description field in the new. It might be the middle name field in the old system converted into the first name field in the new. Incidentally you should be aware that two fields may convert into the one field with the second field overwriting the first. For example, if a first and middle name field are converted into a first name field, you may find that if there was a middle name, it will overwrite the first name, but if there is no middle name, the first name in the new application is correct.

Data which is missing from the conversion and was not part of the conversion plan. This will be missing data because it was never converted. If an old system didn't have provision to store email addresses, and you stored them in a spreadsheet, you may need to convert the spreadsheet. If you are only matching name to name between the spreadsheet and the new application, this might also result in errors with names that are not exactly the same, or where there are two people with the same name.

In order to look at the data you may need to view it through the application. As the application is probably being configured and customised at the same time, there is potential to come across errors that are in fact related to configuration or customisation. If multiple conversions are to take place, it is easier to work with extracts of data from the conversion. For example a list of customers and names dumped to a spreadsheet will probably allow you to see errors.

Another trap will be incorrect data on the current system. When preparing data a decision will have been made as to what data will be corrected before conversion, and what will not be corrected. Once converted you might find more serious implications of incorrect data. There will also probably be data errors you were not aware of in the current application.

The data conversions should continue until there is confidence that a full conversion can be undertaken and provide the database the organisation needs to run their business. That may involve a conversion followed by a series of manual adjustments after conversion. It is tempting to put the difficult conversion areas into this "we will fix it after we run the conversion" category. You should be careful as there is usually only a short window to do the conversion. If there are many hours, or even days, of manual entry after conversion, it can delay the go live timeframe.

Nil  
Nil  

Return to the top