Big data isn’t just for big businesses

By
Anne-Lindsay Beall

More and more midsize businesses are taking a serious look at data visualization. In a recent midmarket survey, 80 percent of respondents agreed that putting data to better use could help them improve product quality, uncover new business opportunities and speed up decision making. 


Ninety-six percent had big data projects either operational or starting up.

The reason for these high numbers is simple. Visualizing your data is crucial in making sense out of the huge amounts of it that can now be tapped. But with limited budgets, limited IT resources and (for the most part) no highly trained data analysts on staff, many midsize companies aren’t sure where to begin. Excerpted from the white paper, How the Midmarket can Take Advantage of Big Data, here are five practical tips for getting started:

1. Build the business case. Vague promises related to improved product quality or better customer service aren’t enough to justify investing in a data visualization solution. If you want to move to data-driven decision making, you need to think through exactly what the business benefits of better data analysis will be, and how much those benefits would be worth. This isn't as complicated as it sounds.
   
   For example, data visualization is highly successful at growing the size of shopping baskets by analyzing previous customer behavior (plus other factors) and proposing the up- and cross-sell items that specific customers are likely to choose. A simple spreadsheet can show the dollar value of a 1 percent increase in basket size, a 2 percent increase, and so on. The same sorts of questions can be posed for any aspect of a business: opera­tions, engineering, human resources, finance and even IT.
   
   What-if scenarios like these are not difficult to calculate, and they put the need for a data visualization solution on a solid business footing.
   
   
2. Collaborate and cooperate. Data visualization is an area where you can’t go it alone. The midmarket survey already cited identified successful collabora­tion between business units and IT as one of the most important success factors in data analytics projects, and lack of cooperation between the two as the most important cause of failure. The message is obvious: If you’re a business manager, you have to get IT on board, and if you’re in IT, you have to sell the business managers.
   
   Another related success factor is obvious but still worth stating: Buy-in from senior management is essential to success.
   
   
3. Democratize your data.  Data visualization solutions were initially developed as a business tool for enterprise-scale companies that could afford to hire statisticians and other data scientists capable of sophisti­cated data analysis. Often, these experts functioned (and still do) as internal consulting groups. This model is too expensive, slow and clumsy for midsize businesses, and should be avoided at all costs. If you’re serious about making data-driven decisions the rule in your organization, you have to make the data upon which decisions are based available without intermediaries – and in a useful form.
   
   This is an area where having the right technology plays a huge role. Data visualization solutions now exist that not only serve the needs of experts, but can also be put to use by nonspecialists. These solutions guide managers through a self-service analytical process.
   
   It’s possible, for example, to systematically analyze data to see which variables are strongly correlated with desired outcomes, or not correlated at all. This eliminates the need for manual trial and error at the beginning of a project to determine what data is relevant. These solutions also simplify the process of communicating insights by suggesting the best way to display data, e.g., with bar charts, pie charts, heat maps or scatter graphs. In other words, they go far beyond the capabilities of spreadsheets, without requiring specialized training.
   
   
4. Ask for help. Don’t let a perceived lack of technical talent stop you. If you have a clear business objective, you can engage consultants on a limited basis to obtain the technical expertise you need to get a data visualization tool up and running, as well as customized training for the user base. This is a much more practical (and economical) approach than trying to hire the talent you need, which may be hard to attract if your business isn’t a giant.
   
   
5. Don’t ignore the need for speed. The speed of a data visualization solution isn’t something that only concerns the IT department. A system’s speed has two very practical business consequences.
   
   The first is that managers who are trying to figure out a problem need a system that works in real time. Problem solving in the business world is an iterative process where each answer leads to the next question. If each answer requires an hour of calcula­tion, it’s very difficult for users to maintain continuity of thought. Managers tend to be men and women of action. They’re likely to abandon a system that requires days of patient waiting to deliver a useful result.
   
   There’s another, more technical, reason why speed counts. A slow system simply can’t process the vast amounts of data now available to midsize companies. The workaround for this problem is to analyze samples rather than the whole data universe. Unfortunately, selecting samples that will accurately represent a larger body of data requires a level of expertise that midsize companies rarely have.