5 keys to data success: Analysis

Mik Data Science April 17, 2016

Analysis provides the final leap from data to the insight you desire.
Analysis provides the final leap from data to the insight you desire.

From data to insight.

The question you’ve been asking through the whole “5 Keys to Data Success” series: how do I actually get from data to some tangible benefit in my company? Thanks for sticking with us as we got to this point! While analysis frequently gets all the credit for the sexy, hard part of data, it’s just one piece of the puzzle. Of course, without it all of your investment in data collection, management, and preparation are for naught.

Data analysis is often the most scary part of a data initiative. Other pieces of the initiative so far have fallen, to greater or lesser extent, in the domain of teams and divisions we are more familiar with existing in companies. Only fairly recently have companies begun to really think about how to directly address the data science in the same way we’ve come to address IT, for instance. But don’t let it frighten you off — you don’t have to hire a bunch of rocket scientists or be an expert in mathematics to succeed. There are a few key questions you should be asking, though, to maximize your investment:

1. What do I hope to understand better from the data?

What’s the metric I hope to hone in on (or even discover)? What problem am I trying to fix through data? Understanding your target is the biggest part of success and keeping costs low. The answer is in you business goals, not an equation.

2. Do I believe the data can provide the understanding I need?

Related to #1. Does my experience lead me to believe analysis can further my business? If it doesn’t, or you can’t get buy in from other stakeholders, you’ll never integrate the resulting insights into your business process — making it worthless!

3. What type of analysis do I need, and who will perform it?

Need artificial intelligence or just simple visualizations? This informs who does the analysis: data scientist, company analyst, your assistant? Consider how this fits into workflow, if adding to existing roles, or how dedicated roles will interface.

4. What is the result of the analysis, and who needs to use it?

Be specific. Aligning the form of the results with those who have to use it ensures the insights are actionable and understandable. Connecting the technical side of data analysis to the real-world application is hard, and best addressed upfront.

5. What’s my budget (seriously, this time)?

Analysis often comes with high upfront and long-term costs. Adjust expectations or revisit the questions above to ensure you are in budget. Data scientists cost a lot ($95k/year starting salary; $300-$450+/hour consulting), so know your needs!

Analysis is often seen as the most costly step. This is due, in part, to the high initial cost to get started with data scientists as well as the perceived lack of benefit when analysis results aren’t readily actionable. Considering where the analysis workflow and outcomes fit in your business from the start will ensure proper integration of the insights — and a better ROI. The questions above are just a few to get you thinking about what matters in the big picture. Data analysis should be about your business, not about technology!