So, you’re thinking about investing in Apache HadoopⓇ (or related Big Data technology). Clearly you recognize that in the modern digital economy, enterprises don’t just compete based on the quality of their products, but on their use of data to continually improve and service those products. Looking towards the future, data is driving the next generation of compelling, personalized customer experiences. Of course, many examples of ROI from Big Data already exist.
As GE CEO Jeff Immelt put it, “If you went to bed last night as an industrial company, you’re going to wake up today as a software and analytics company.” From our perspective, just replace industrial company with whatever industry you’re in and, I assure you, the statement applies—automotive, education, media, insurance, and many others.
Technology is just one part of the Big Data equation, however. There are a number of less tangible but equally important considerations that will determine the outcome of any Big Data project. So, before you open up your wallet, make sure you can answer the following questions:
1. What is the business use case for your investment?
There is nothing sadder than a shiny new Hadoop cluster (or other Big Data technology deployment) sitting idle in the back of your data center. But, that’s precisely what you’ll end up with if you don’t clearly articulate a business use case for your Big Data technology investments ahead of time. But don’t pick just any old use case. Find one that is manageable but also core to your business. Don’t think of the project as a proof-of-concept that you just scrap when it’s over, but rather a real project that will deliver real, ongoing results for the business. If you work for a financial institution, this might mean tackling fraudulent credit card activity. For a retailer, it could be personalized customer offers. Clearly lay out the business requirements and aim for a quick win.
2. How will you measure success?
Speaking of quick wins, when laying out the business requirements of your initial Big Data use cases, make sure you clearly identify the metrics you’ll use to determine success (or failure). In other words, you have to know how you are performing today in order to show improvement tomorrow. For example, a shipping company embarking on a Big Data project to optimize delivery routes must know how many miles-per-gallon it’s fleet is getting before the project starts to measure against miles-per-gallon upon completion of the project. If you don’t have a baseline from which to start, it is next to impossible to gauge results. And success breeds success, so it is important to show measurable value with your initial Big Data use case to build momentum for future projects.
3. Who is leading the initiative, the business or IT?
If the answer is the latter, do not pass go, do not collect $200. Now, don’t get me wrong. IT involvement in any Big Data project is critical. The technologies involved can be complex, and IT is often best equipped to implement and administer them. But, even if you’ve identified a compelling initial use case, success is far from guaranteed without buy-in from the business. The better approach is to get all relevant stakeholders—IT and the business—involved at the start and identify a business owner to champion the project to colleagues and management. This will increase the odds of success. If you’re in IT and can’t find a business sponsor for your first Big Data project, maybe your business case isn’t as compelling as you thought.
4. How do you plan to address the Big Data skills gap?
There’s a well-known shortage of data science and analytics professionals in the market, meaning your enterprise likely doesn’t have a team of all-star data scientists sitting around just waiting for you to bring them an exciting project. So, you need a plan to make up for this skills gap to actually do the hard data science work required to get value from your Big Data technology investment. It is a good idea to find a partner with data science experience in your industry, as well as one that will train your team on the discipline of data science over the course of the project—better yet, find one that does pair programming to ensure hands-on skills are transferred. Among many others, the Pivotal data science team, for example, works side-by-side with our clients on-site building data models and algorithms. This way, teams gain important data science skills that can be applied to future projects.
5. How do you plan to overcome institutional resistance?
Getting people to change the way they do things is extremely challenging. This is especially true at enterprises with long-entrenched business processes and an over-reliance on intuitive, “gut-feel” decision-making. Be prepared to lobby aggressively for why data-driven decision-making is required and use examples of similar enterprises that have successfully embarked on Big Data projects. Be persistent, but also be tactful. Nobody likes to be told “the way we’ve always done things” is wrong, and some people will even feel their value to the enterprise is threatened by Big Data and data-driven decision making. Don’t underestimate the change management challenge associated with Big Data.
So I’m glad you’re ready to embrace Big Data, but make sure you can answer these five questions before you make the leap. You (and your wallet) will be glad you did.
- Read more articles from Pivotal’s Big Data team
- Check out our 5 Big Data Predictions for 2016
- Find out about Pivotal Big Data Suite
Editor’s Note: Apache, Apache Hadoop, and Hadoop are either registered trademarks or trademarks of the Apache Software Foundation in the United States and/or other countries.