Friday, December 23, 2005
data quality for enterprise usability
Data management has always been a massive topic in IT, and that shows no sign of abating. Companies often proclaim that data is the key to being customer-centric. Sometimes they mean having data available about a specific customer transaction while speaking to a customer. Other times customer-centric means mining data to predict what customers will do in the future. A good example of both these dimensions is insurance. Data collected from current policies and claims are important for the resolution of issues to the customer's satisfaction. And historical data on past policies and claims can be used to predict customer behavior.
The problem for companies is that while they collect volumes of data, it is not always useful. One study claims that data quality problems cost businesses $600 billion annually. While that figure sounds exaggerated, one reasonably can assume data quality is costly to business.
Usability can play many roles in improving data quality. It can improve data labeling and taxonomies to enable better sharing and aggregation of data. It can explore how to streamline the collection of data by employees and from customers. It can map touchpoints where data can be verified from customers easily, to allow data such as addresses to be updated and corrected. It can improve retrieval and analysis of data, for example through drill down techniques, so it more often sees the light of day. Most data collected by companies is never used again a week after it was collected.
In short, usability can help with the accuracy, completeness and relevance of data. A fair amount of data collection and analysis is automated, and usability has little to offer those processes. But if the automation worked as well as it is supposed to, data quality wouldn't be a problem. It always comes back to people.