Companies are finding that it’s a lot more challenging to implement big data than they thought it would be! And some are losing enthusiasm.
What are ten things you can do to prevent a loss of momentum in your big data projects?
1. Keep pilot projects short—and make them a success!
This is a fundamental for salesmanship and project management. You should always keep pilot projects for new technology concepts short and ensure that the proofs of concept that you apply to them have a high probability for success! This rule of thumb is no different for big data than it is for any other project.
2. Make your pilot project show a business path.
This is an element many sites don’t even think about in pilot project work—especially with big data. They should. Your goal is to show how big data can deliver tangible results to the end business over time—hopefully, a long time. You can economize your pilot project efforts if you also ensure that there is leverage built into the pilot project’s results. In other words, if your goal is to show analytics gathered from how your customers are using your ecommerce website so you can increase your revenue capture, it might also make sense to “stub in” enough common logic (and data) so you can go out and capture analytics from social media sites, in subsequent projects phases also. If your goal is to track Internet of Things (IoT) data that tells you if your trains are running on schedule and if tracks are in good order, you might want to schedule a phase two project that additionally automates maintenance scheduling and expediting with your track crews so they can fix a track before it becomes a hazard.
3. Use metrics—and demand business results from them.
It’s great to measure system throughput and mean time to result for big data from an IT standpoint—but if you don’t have high impact metrics and results from big data for the business itself, the support of end business users is going to wane.
4. Have champions in the business.
Big data initiatives are only going to thrive long-term if they deliver tangible value to the end business. Consequently, big data is not an initiative that IT should pursue in isolation. If you don’t have your business people on board and sufficiently informed so that they know what they should expect from big data and what it is going to take from both the business and the technology sides of the house to get there—you’re better off waiting until you have the groundswell of support you need from the business.
5. Have C-level visibility.
Big data should deliver results that can impact the corporate bottom line through savings or through new revenue opportunities. In other cases, it might deliver significant strategic advantages. But if it’s not doing any of these, it’s not going to capture the attention of C-level executives for long. Stay focused on the end business results of what you’re doing—and communicate progress regularly to your C-level executives and their managers. Communications from IT should always be in plain English (i.e., no technical jargon).
6. Attain organizational support for the build-out of the entire big data pipeline.
The old axiom “garbage in garbage out” (GIGO) doesn’t change with big data. One of the real “drag points” for big data projects is to clean data up so that it is of high quality before you begin to plumb and analyze it. This is where the CIO should call a meeting of high-level executives to explain the end to end big data “pipeline”—and also the necessity of mundane operations like isolating and cleaning up data before it is plugged into analytics processes. People are more tolerant of what they initially might perceive as “non-value-added timewasters” (like cleaning data!) if they understand how the upfront grunt work is going to contribute to better results down the line.
7. Identify the right ROI goals.
Because they have little experience with big data, many sites begin justifying IT investments in big data by talking about data throughput and processing in the same way that they would talk about such things in the transaction processing environment. True, one goal of big data is to process data quickly for fast results—but a large part of big data investment returns are in the way that parallel processing maximizes server use. Server utilization for big data can approach 90-100 percent. In contrast, transaction processing servers might only be active at 20-30 percent of capacity. This is what IT should be talking about when it comes to big data return on investment.
8. Develop a coexisting strategy with existing reporting.
It’s important for everyone to know that the intent of big data is not to replace reporting systems that already are in place and that have been returning value to the organization for years. Rather, the goal of big data is to tackle new questions that the organization was never able to answer with legacy reporting. It is the strength of these two reporting strategies together that will deliver on the organization’s information assets.
9. Set up a big data team.
Big data can’t be approached in the customary ways that companies have always developed reporting. This means that the traditional end business users and IT who have teamed in the past on corporate reporting might not be the same people who are suited to the harvesting of big data. A big data team in most cases minimally requires an astute business analyst who can work with the end business so the right analytics queries are formed; a savvy data architect who can serve as the DBA so that the appropriate big data repositories are created; and a data scientist who knows the procedural and non-procedural programming tools that can be used to mine the data for analytics information. This team often ends up reporting through IT, but it also can have dotted line relationships with end business units.
10. Get help in question formation if you need it.
The number one stumbling block for many companies that begin to sour on big data is knowing which questions to ask of their data, and how to develop questions. If you feel that the organization’s wheels are getting stuck in this rut, don’t delay. Seek an outside business analytics specialist or consultant who can come in to facilitate the development of appropriate analytics queries that can get to the bottom of what your business wants to know—and also help transfer these skills to your own people.