Avoiding the Disappointment of Data
Are you getting the greatest value from your data? Here are five ways to make sure.
- By Patrick Siconolfi
- September 28, 2018
When I speak to business owners or executives, audience members never say, "I love the insights I get from my data" or "Data has helped my company grow by X percent or make $X." Rather, I tend to hear, "It takes forever to get the information I need to make a decision," or "I've spent so much on data but I don't see a tangible return."
With all the great things data has done for businesses in the last 10 to 20 years, why are so many people coming away disappointed?
According to a recent article in the Harvard Business Review, on average:
- Less than half of a company's structured data is actively used
- Less than 1 percent of a company's unstructured data is used
- More than 70 percent of employees have access to data they shouldn't have
- About 80 percent of analysts' time is spent simply preparing data
Given these statistics, it's no wonder people are disappointed. Companies are buying into the promise of data and investing millions -- so where's the return on their investment?
The disappointment comes down to one thing: a disconnect between strategy and execution. The people defining data strategies (consultants and executives) have never actually executed on the strategies they advocate. That's the responsibility of technical resources (developers, report writers, data analysts) who rarely understand the strategy, let alone the business purpose it's meant to support.
5 Ways to Restore Your Faith in Data
What can you do to avoid data disappointment? The successful data projects I've been part of have been rooted in foundational business objectives. Here are five key strategies I've learned over the years.
1. Figure out "what" and "why" first, then focus on "how"
Starting a project by picking the technology first and then deciding how to bring disparate data together has never been a recipe for success. Give your project a defined outcome that's tied to business value. Focus specifically on the goals the data project is meant to support (the "what" and the "why"). Only then decide on the "how" -- including the technology you're going to use.
2. Deliver value quickly; think small and agile
Use agile methodologies in your project; it's the best way to deliver value quickly. This way, business stakeholders can validate that things are headed in the right direction, and you can change course quickly if they're not.
Large project teams with long project schedules can't offer the kind of speed you need. Smaller agile teams with experienced consultants reduce project management overhead and communication gaps. They typically have multiple skill sets; a data modeler builds the ETL routines versus a data modeler documenting the required ETL rules and instructing the ETL developer on execution. This can reduce the communication and documentation overhead found in larger teams.
Small teams can build something themselves and check in with stakeholders as they go. A larger group documents the process first, then hands it over to developers who may not fully understand it, with no room for frequent iterations or back-and-forth with stakeholders during the project. As a result, stakeholders often end up with something that's not an exact fit for their needs. On the other hand, when results are delivered quickly and validated by key business stakeholders at each stage, there's far less risk that the results won't meet expectations.
3. Focus on the data that will solve specific business problems
What questions do you and your business stakeholders want answered? Define them first. The solution can be crafted to store and manage data in a way that facilitates the answers. Determining what data you really need is more efficient than trying to make sense of all your company's data at once.
Typically, you'll find 80 percent of the value in 20 percent of your data. Teams that can laser focus on the data that provides the greatest value will ultimately be more successful. Their efforts won't be diluted by irrelevant or low-value data.
It's easy for a team to get caught up trying to deliver on 100 percent of the data. Start by establishing value from the first 20 percent (the most essential), then move on to the next, less-critical 20 percent, and so on.
4. Work with the right people
Achieving a strong data strategy requires bringing in people with a track record of success with these types of initiatives. All too often, organizations view data and analytics as an extension of other IT initiatives, or they lean on resources who have delivered applications or infrastructure projects but have no real data and analytics project experience.
The best people to mine your data for the highest value may not be found in your technology department or in a big consulting firm. Instead, look for well-rounded, technically skilled people with strong communication skills and the ability to understand the needs of the business stakeholders driving the project, then apply that understanding to data sets.
Don't look to offshore resources; the people you need are typically not inexpensive but they can deliver more value with less effort and in less time than anyone else.
5. Put data governance and access processes and technology in place
To get data fast, governance and controls don't have to take a back seat. Without controls, you're using data that's compromised, and your results will be compromised as well. This is where many projects fail to meet business objectives.
A common scenario that I see involves people accessing raw data through Excel or data exploration tools such as Tableau and Power BI without any controls behind the scenes. We are seeing more of this because BI tools have become so powerful and flexible that they empower almost anyone to build their own analytical view of the organization. The results lack accuracy and consistency; there's nothing stopping users from applying their own lens and rules to the data and building their own analytical view.
Governance doesn't need to add cost and cause delays. By using technologies that promote governance by using a common metadata repository governing all data/business rules, you minimize the amount of process and organizational change required to achieve governance -- and end up with dependable data you can use to solve your business problems.
Now Is an Exciting Time to Leverage Data
We are experiencing an information revolution, and like any revolution, it brings change. Companies have a chance to truly differentiate themselves from their competitors and create products and services they wouldn't have dreamed possible just a few years ago. At the same time, revolution also introduces fear -- fear of the unknown, fear of change. I think it's a matter of looking at the glass as half empty or half full. Which way does your company see things? What's the status of your glass?
Use data properly and it won't disappoint. Plan carefully, invest wisely, and you'll be poised to reap the rewards and achieve success.
Patrick Siconolfi is a co-founder at Gensquared where he is responsible for leading the data integration practice and analytics cloud platform. You can reach the author via email or on LinkedIn.