A few years ago I came across an infographic representing most of the vendors in the “big data landscape.” It contained hundreds of companies, some of which had been well established for a long time, some of which were acquired by larger companies, some of which went public, and some that inevitably went out of business.
It got me thinking—how does anyone (outside of the analyst community) begin to make sense of it all? How should an organization approach making decisions about which solutions to invest in? Will large product suites fill the needs of those organizations, or are best of breed technologies more appropriate for certain problems?
As it pertains to data strategy, I believe that such decisions should start with an understanding of the organization’s current data landscape, hypotheses about where value can be added (and how much value), and planning to tackle problem areas and advance organizational capabilities.
Assessment - where do we stand today?
It’s important to assess where your organization stands currently, along many axes:
- Data sources - What are they? Which are being used for reporting, analytics, and/or predictive? Which ones are not?
- Data architecture - Do you have an operational data store, data warehouse, and/or data marts in place currently?
- Data integration - Are there established tools, processes, and automated jobs in place to prepare data for use in decision making at various levels?
- Reporting - What tools are in place today that support pixel perfect reporting and delivery?
- Ad-hoc / self service analytics - what tools are in place today in support of true self service analytics?
- Extended data services - are any data assets currently being extended as a service to partners, suppliers, or customers today?
- Security and governance - what aspects of your enterprise data are being secured, at what part of the overall architecture, and how are these controls managed?
- Predictive analytics - what tools are in place today in support of predictive analytics? What kind of algorithms are being leveraged? What insights are being surfaced, and how truly useful are those insights?
By assessing each of these areas, you can get a pretty clear picture of where your organization’s current gaps are, where there is pain in the process for various stakeholders, and where there is untapped value. Sample questions to consider for these various areas of enterprise data strategy:
- Are there any delays in the processes surrounding data driven decision making?
- What are the potential costs of those delays?
- Are there any staffing deficiencies causing certain bottlenecks?
- Are there any unnecessary “science projects” that lack clear direction?
- Are there untapped data sources that could yield valuable insights for the business?
These questions form the basis for a “bottoms up” analysis. Conversely, a “top down” analysis can also be useful in hypothesis generation. Ask yourself questions along these lines from the business perspective:
- What decisions are being made by the executive team on a daily, weekly, monthly, quarterly, or yearly basis? What data is needed to make those decisions effectively? These questions should be asked of senior leadership, all levels of management, and all users of enterprise data both internal and external.
- What is the potential business value, in dollar terms, of the decisions being made based on data?
- What is the potential business value, in dollar terms, of the decisions that are not yet being made based on data, but are currently being made based on gut feel alone?
Between your “bottom up” and “top down” analysis, the answers to those questions are likely to provide good insight in the highest leverage projects that could be tackled to provide the most business value, whether they be in pursuit of cost takeout or revenue creation. Prioritize that list of hypotheses in terms of potential business value in descending order, and then evaluate each in terms of complexity and time to implement. Identify what “quick wins” can be tackled, and which projects are longer term initiatives that could pay off significantly.
Plan and execute
Once you have your list of hypotheses prioritized, determine how to parallelize your efforts. What are the next steps for the quick win projects? What about the longer term, more strategic initiatives? Figure out how to move the ball forward, and how to marshall resources in support of multiple objectives at once.
But never forget that you don’t know what you don’t know. Go to an industry conference or two, or send members of your team to do so. Find out what is emerging in the industry, and leave room for innovation in your enterprise data portfolio. Your product suite vendor is not the most likely place to find bleeding edge innovation—go find out what else is out there, and factor in solutions that add value you didn’t previously anticipate.
Maintain and measure
As you execute on your high value projects, go back to the hypotheses you generated and track progress against the anticipated cost savings or revenue generation. Is what you’re finding in line with what you estimated originally? Why or why not? Identify what caused any variation, and remediate as appropriate.
In summary, though the vendor landscape is ever evolving and difficult to keep track of, your solutions should be in support of your organization’s data strategy. Take the time to assess the current state of your enterprise data management capabilities, identify gaps, pain points, and opportunities, and hypothesize about how to advance organizational capabilities to support cost removal or revenue creation. Finally, plan, execute, and measure progress along the way, and always leave room for innovative solutions!