Want to profit from Hadoop? Consider these 4 reasons for developing a big data innovation lab


42-61534963How can you use an innovation lab to be as agile and innovative as a startup? Are there different types of innovation labs and if so what is the difference?

I answered these two questions in previous posts, and now I will answer a third pressing question: how can you build the business case once you've decide to develop an innovation lab?

Providing concrete numbers is hard, as it varies from organization to organization, but there are some clear areas to research as you build out your proposal. In this post I cover four compelling arguments for establishing a dedicated platform and services for innovating with big data. My hope is that this list will help support you in building your business case for creating an innovation lab, so that you can profit from Hadoop, big data and analytics.

1. Too much time and too many resources are spent trying to justify one-off big data project investments.

As I eluded to in my previous innovation lab post, it can take an army of people across an organization a great deal of time to outline, justify, approve, deploy and try out an idea related to big data. This is especially true when using traditional approaches  for proving out IT software and building a business case for investments.

These slow, cumbersome processes can lead directly to slower time to market, at a time when the digital world is accelerating. Likewise, the people dimension is important since it is a bottom line cost that most organizations are trying to keep under tight control. Those first two points come on top of the hardware and software costs organizations would incur for trying something new which they might never actually deploy.

When you combine all those factors, it's clear an innovation lab can derive significant cost savings in terms of people’s time (and therefore cost). When used correctly, the innovation lab should significantly reduce the number of business cases that are built but never acted on or end up being compiled and executed before being discarded as they did not deliver to expectations after a great deal of effort and money is wasted.

By optimizing the process of big data projects, a great deal of time can be saved which leads to first mover advantage, normally helping to drive the top line (the value of which is hard to quantify). Beyond that, when you put into place a consistent platform to support your big data efforts, your costs remain largely static over time as ideas are tried on the hardware and software side.

2. Adjusting operational systems and processes because you “have to do big data” can be expensive and may not deliver the results you expect.

Most organizations do not have the luxury of starting from scratch when developing their next generation IT architectures. To exploit big data quickly, organizations need to make a plan for getting results within the current operational environment, including a plan for how much to expand or tweak those environments.

Typical infrastructure questions we see often sound something like this:

  • If I join network data and customer data together (which can be petabytes of data), can I get reduced churn as compared to just using customer data?
  • If I include weather factors with my life insurance risks can I better predict the probability of death to more accurately price life insurance premiums?
  • If a local venue brings in sporting events can I determine whether the events will affect network reliability or increase dropped calls in the area?

To test these types of questions, you need to bring together a lot of data, and try it. Often, the most valuable data in that experiment will be data you do not already have. In the old days, we would build the business case around the potential reduction of risk exposure or the churn reduction and present the profitability impact. Based on that, we would make the needed investments, maybe in a larger DBMS, for example, with the goal of getting a result that would outweigh that investment. Sometimes the cost of the infrastructure changes would skew that business case and we would not proceed to even try it. (Adding multiple terabytes of storage and CPU processing capacity to a database can be very expensive, for example.)

The innovation lab provides you with an environment to test out your hypotheses and determine if they are true or false BEFORE you try to build the comprehensive business case. Of course, neither outcome is inherently bad as a false hypothesis tells you to stop investing, which is a money saver.

The innovation lab also helps identify which data will be valuable to bring together in existing systems (and which to exclude), thus limiting the bloating and related costs of existing systems. The pilot will also provide a solid initial analytical model, if that is what the lab is trying to develop, that can be refined and deployed.

In this sense, the innovation lab is going to help you refine the project. It will help identify which specific data needs to be added to what you have, which is often a subset of everything that is available. The innovation lab will help you know which models you are most likely to use and help create the initial versions. Then you can refine them and deploy them into your operational systems as opposed to starting from scratch once you have been through the lab.

Finally, the lab will tell you up front the likely improvements you will see so that the business case is proven before you start, meaning no more great big data hopes that turn into nothing messing with your operational systems.

In short – the innovation lab will provide you with a proven business case and costing before you begin, or it will notify you to stop what you are doing before you waste a lot of time and resources.

3. Overcoming the skills issue and associated ongoing costs of big data.

As we know, there is a shortage of skilled people who can work with the combination of big data technologies we see in play today. What this means is that the entry price to get going with big data is, at the very least, high  training costs and perhaps a number of additional staffing costs for employees who can focus on the new technologies.

Today, these "data scientists" are in scarce supply unless you are Google. An innovation lab approach will provide a central place to locate appropriate staff so the whole organization can experiment with big data with the goal of transitioning the results into the operations as appropriate ensuring the efficent and optimal utilization of resources.

The truth is, most people on the business side of your organization do not need to care about big data, as they just want better, quicker answers. By having trained staff working alongside those with a detailed knowledge of the business, and the data, the innovation lab will help to nurture repeatable best practices for testing out new hypothesis incrementally and lowering the cost of such projects over time.

At the same time, ideas coming from one side of the business can be shared with others enabling a virtual collaboration to happen. All in all, the innovation lab should reduce the cost of supporting big data projects and significantly reduce the time it takes to complete them by helping to centralize the capabilities within a supported service driven approach on a standard known platform.

4. Agility without compromise.

Disruption is now rife within almost every industry. If you are not disrupting proactively within your organization, it is likely that you will be driven to change by external market disruption (i.e. new market approaches taken by your existing competitors) or by some disruptive new entrant into the market. Disrupt yourself or risk your company going out of business.

In my first blog post about innovation labs, from October last year, I spoke about having the ability to innovate like a start-up. The innovation lab provides that platform to try and beat your own company from within, a place where you can depart from the norms and try out radical new approaches and ideas in a segregated environment to see if you can be the disruptor and not the disrupted.

It is very hard to put a value on this as it goes to the core of many businesses. A new approach internally could lead to dramatic productivity savings, open up new lines of business or new ways of driving value from your data that could deliver great shareholder value (think data monetization here). A new approach externally could redefine the services on the market differentiating you from all your competitors and making you a leader in your market, which again often drives significant shareholder value. In short, the innovation lab could be the thing that helps save your organization or elevates it to a goliath of the industry!

Wrap up

I believe passionately that big data offers a chance for organizations to do things differently but I also see a lot of potential for huge spend with little return. An innovation lab provides a platform on which a programmatic approach to gaining value from big data can be put into place. I think it will pay for itself many times over within the first few projects in savings or improvements to the way of working. Longer term, it could be the engine that keeps your organization running and establishes you as the dominent market leader.

For existing SAS customers, the idea of a lab provides an ideal place for them to benefit while also getting to grips with the capabilities of the modern analytics platform of SAS as well.

Let me know your thoughts and feel free to follow me on Twitter @mark_torr to see what else I am passionate about.


About Author

Mark Torr

Back to Top