Selling big data projects requires selling big data ROI. Unfortunately, since big data is still in its infancy, making a case for big data ROI is a lot like making the case for social media when it was still a new concept – you know it pays off, but it’s difficult to show how or why.
Despite the fact that measuring big data ROI is difficult, companies are still optimistic. A study by the AIIM Market Intelligence Division on measuring the ROI of big data and content analytics revealed:
- 62 percent of users said they would find analytics “very valuable.”
- 20 percent said they would find analytics “hugely valuable,” especially when linking structured and unstructured data together.
- Those surveyed see the greatest big data ROI in improving data quality, addressing policy compliance, and speeding up customer service.
- While 56 percent are feeding big data results into decision-making, only 6 percent are using big data for strategy and 7 percent of big data projects reported unsatisfactory results, probably because they didn’t know how to apply the findings.
The best way to determine if you are getting the right returns from any big data initiative is to set the proper parameters in advance, and then measure as you go. Big data is not an information oracle designed to answer all your questions. You need to start by asking a specific question, then gather the right data, develop a methodology and model for analytics and analysis, in order to assess the findings.
Two Paths to Big Data ROI
When considering how to gauge ROI from big data, you basically have two approaches to see measurable returns:
- Assess the cost of big data projects against alternative approaches. Consider, for example, that you are considering expanding your data warehouse to accommodate more and different types of data. You could use OLTP technology and add more database servers and software at a cost of millions of dollars, or you could achieve the same result using a Hadoop framework for about $200,000. You will still extract value from the existing data in the warehouse, but the Hadoop infrastructure relieves the need for added data processing capacity.
- Measure the value of big data insight. It’s difficult to measure the value of insight, but you should start by limiting the scope of the big data project. It’s a lot like mining for gold. You start by sampling the data to see how rich it is before you commit to big data mining. If you start with a proof of concept, you can keep the scale of the big data project small, using cloud storage and resources and less expensive tools and hardware to determine if you will get valuable results. If the results from the proof of concept are promising, you can move to the next stage, committing more resources, more analytics and software, and more money.
Measure Time-to-Insight as Well as ROI
The three greatest obstacles to measuring big data ROI are:
- Understanding how to get real value from the data;
- Properly defining the scope of the big data investigation; and
- Finding the right resources and skills to extract insight from the findings.
The challenge is finding metrics that let you measure big data ROI in an exact way, and more importantly, to understand how soon to expect a return on big data. A recent report from IDG Research and Kapow Software notes that big data projects often take longer than expected, and they usually return less ROI than expected because they require outside consulting expertise to extract meaningful insights. Be sure to consider time-to-insight and the need for outside assistance as factors when gauging big data ROI.
The best strategy to increase ROI and reduce time-to-insight is to start with smaller big data projects. In other words, use big data as an enabling technology rather than as a solution in itself. Limit the scope of the queries to reveal answers to questions that can yield immediate returns.
For example, one company used big data to learn more about customer behavior to convert one-time purchasers into repeat customers. The scope of the project was focused, and the ROI was a 120 percent increase in sales. The time-to-revenue was shorter, and the proof of concept was sufficiently promising that the company started working on larger big data engagements requiring more technology and consultants to analyze the results.
As with any project, the way to help customers understand ROI is by setting proper expectations at the outset. If you can limit the scope of a big data pilot project in order to prove ROI and demonstrate the potential returns from big data, it will be easier to sell larger projects with bigger returns. How do you help customers understand big data ROI?