In order to provide the best possible service it is important that data projects are understood. Like all projects, it needs to be to budget, to timescales and to the desired quality. From our many projects the following key guidelines are important to know when planning a data project:
1. Data Project Timescales Depend on the Quality of the Initial Data
The initial quality of any data will determine how long a project is and what the quality of the end result can be. For example, in address management, if an address has too many elements missing then it is very unlikely you will find the correct address through a standard address cleansing procedure. You may need to search for the address, you may need to call the owner of the address, or you may use other information such as company details. In each of these cases, the workload has increased because the initial data was poor.
Poor data quality will lead to poor results for any automated system, leaving a great deal of manual work for everyone. Often temporary staff are hired because standard software isn’t good enough to solve the problem and your desired quality isn’t being met.
Knowing the initial quality will give you an insight into how much work is involved and how much it may cost. Looking after your data regularly is the key to keeping long term costs to a minimum.
2. Apply the 80/20 Rule to Resolving Data Issues
If you determine your data quality to be 30% good and after you have completed the project you calculate the quality to be 90%, then you have a 300% improvement! Focusing on the remaining 10% will give you diminishing returns in terms of money and time spent. Too often we look at every little error and forget the bigger picture. Often with data it is not worth pursuing every small change that is required. You must decide prior to any data activity what level of uplift you require and remember that the last few percentage point improvements will take the longest.
Aligning the 80/20 rule to what the business really needs will give you the maximum results for the lowest cost.
3. Client Input is Essential
Sending data to a data supplier is ok when it’s a bureau service, where an automated system cleanses the data to some level. Data can always be improved further if you need it to be. By pre-processing the data beforehand and getting the client's input you can get significant improvements. Client involvement is essential when identifying duplicates and golden records and migrating data. Their knowledge is essential to making those important decisions about your data.
Deduplication, Address Cleansing, Golden Record Identification, Suppression Matching and Merging all require client input to get a real result.
If you would like further information on the 3 key guidelines for a successful data project contact Acuate:
|Telephone:||+44 (0) 20 8799 0360|