Mo Data Mo Problems?

With the best intentions in the world, as you collect more data it will eventually lead to problems. Before you get to that point there is a journey of getting more data from more data sources. The more planning and structure that you have in that process the longer you can hold off that impact.

Data steps:

Have an overall strategy in mind; who is going to access data, who needs it, what do they need
Understand the potential for integration; not every dataset can or has to be integrated. Integrating will provide a clean consistent view – however, there are also gaps i.e. datasets rarely overlap
Plan for mistakes; there will be errors, there will be mistakes. Some systems allow you to roll back, others require you to start again. You need a plan for each.
Governance; have some overall arching rules on how and where data is going to be processed and stored, and how this relates to other sources.
Implementation; do one thing, do it well, then move on. It needs to be right but there is no point continually reinventing the wheel
Consistency; this is perhaps the most important, data sets are built up over years if you constantly change the structure or level of data you are capturing you cannot use the history so it becomes useless

Think you’re a master of the data universe?

For those who confidently say that data quality is not an issue for their organisation here are a few ways to test your data sphere:

Ask an ambiguous question

For a retailer, you could ask different teams how much profit you have made this quarter.

A marketer may take all sales-cost

The merchandise team may take sales-returns-cost

Finance may take sales-returns-cost-contribution

What would your teams say?

Reconcile across systems

Does each of your systems show the same number of actions?

compare categories

For each system compare categories and groupings, can you match one category to another across systems?

Still smiling?

I hope so. Otherwise, I look forward to hearing from you soon…