Data Wrangling

Companies are awash in data – financial data, sales and marketing data, employment data, customer data, partner data and more.  So how best to wrangle the data for meaningful insights?  We’ve hosted a few ‘big data’ sessions at the MIT CFO Summit over the years, and from that, below are some top data do’s and don’ts.

Specify the problem you’re trying to solve.   Make sure you identify the problem or issue you are trying to solve at the beginning – what specifically do you want the data to discover.

Start small.  It’s best to start small – for example, consider a data set that contains five years of numbers rather than 15.

Make sure the data is consistent.  Data that is collected in one place may be monthly and in other places it may be weekly.  It may have been collected on a days-ending basis or on an average basis. It may come in different formats: PDFs, databases and/or websites. It’s important to utilize a uniform universe.

Pressure-test the integrity of the output.  Once you’ve run the numbers, line up the analytics with your real-world knowledge.  Do the findings  make sense? Take the time to assess and evaluate the conclusions.

Diving into data can sometimes seem overwhelming. But with a clear objective and “take charge” commitment to both reliable input and output, businesses are deploying new and continually expanding ways for integrating analytics into day-to-day business decision-making and success.