In the age of “Big Data,” an increasingly common question is whether there’s any benefit to small data, especially for companies with already running business intelligence efforts. The answer, of course, is absolutely.
Data of all volumes are useful, just in different ways. Small data solves problems of equal value to big data, and can make big data analysis easier.
Small data, in this instance, refers to a data pool that contains a relatively lower volume of entries or sources than would be expected from “big data.”
In general, data stores that fall under this definition are light enough to fit on Excel (one million rows by 16,000 columns) and composed of either a single or a few different data sets. This is small enough to reside on, and be analyzed by a single computer.
This leads into Allen Bonde’s popular definition, “Small data connects people with timely, meaningful insights (derived from big data and/or ‘local’ sources), organized and packaged – often visually – to be accessible, understandable, and actionable for everyday tasks.”
Summary: small data is concise enough to help an average person make a quick, intelligent decision.
But is that useful? In the face of the power and might demonstrated by big data, why play around with a comparatively miniature data set?
Below are four occasions where small data is not only a useful tool, but the optimal tool.
Small Business Insight
The first case where small data is preferable is when big data is simply out of reach. Small businesses may theoretically experience benefits from tapping into an all-encompassing data set, but since data generated by their own company is likely limited, it’s more helpful to know how to gain insight from small, relevant data sets that address their immediate needs.
And really, there’s no good reason to default to the premise that big data is inherently superior to small data. It is more important for analytics that multiple, disparate data sources are connected and correlated than that any particular data source is extremely large. A focused business question, and a plan to find the data-derived answer, is simultaneously cheaper and more valuable for a small business than a team of data scientists managing servers and running tests in hopes of uncovering new insight.
Proof of Concept
For organizations looking to use new big data sets and try out new data analytics tools, it helps to first simulate the analysis on a smaller data set. The first step in any big data effort needs to be identifying a real question or issue for which data analysis may help make a better business decision. If no such insight can be gleaned from a smaller foundational set, then maybe you need to rethink how you plan to analyze the larger set.
Testing Data Features
Small data sets are easier to manage and manipulate when determining which data “features” you need to track to find your intended insight. Quick trial and error would be ridiculous to attempt on the full data pool, due to time constraints and network bandwidth. If your business intelligence method involves complex SQL programming, performing analysis on the small data set could also serve as a practice step.
CEOs and Executive Accessibility
CEOs and executives are becoming increasingly hands-on in their companies’ business intelligence efforts, as demonstrated by the shift towards self-service business intelligence tools. Small or summarized data sets are easier for them (or anyone who’s not a trained data scientist) to explore, than working through the full data set. Once the CEO observes a new angle or feature to look into, the data team can take over and run the model on the big data.
So, yes. Small data is useful. It’s useful whether you have big data or not, and the ability to get insight from a limited set of data remains necessary. Perhaps it’s even the new competitive edge.
Comments, questions? Dissensions, mentions? Comment below or bring it to social media to discuss with your peers.