Big Data Fail

For 70% of companies, ‘Big Data’ fail to generate returns above the cost of capital. This is bad, doubly so because the data is from the telecommunications industry which started exploiting big data early and where the scope for exploiting data is perhaps greater (and surely more obvious) than in most other established sectors. Indeed, by 2014 leading telcos were reporting fully 20% of their profits came from big data, clearly demonstrating the opportunities for getting it right. Let’s examine this in more detail.

The data quoted comes from a peer-reviewed article Reaping the benefits of big data in telecom published in 2016 and based primarily on data from a survey conducted in December 2014 of 273 executives. At the headline level, it is in line with our experiences.

Leveraging data and scientific marketing for commercial advantage has been a strategic imperative for mobile telecommunication companies for decades. The delivery of this advantage took formal shape with the Customer Value Management (CVM) approach which we first created for Vodafone in the early 2000s (see our original case study). We have seen across nearly a dozen operators that full implementation of CVM generates directly attributable incremental EBITDA growth of around 5% per year. In revenue terms, operators typically measure 10% additional revenues in the first year alone.

Data is clearly important in this industry and the value is clearly demonstrated. So what goes wrong with the big data projects?

The article provides a list of issues; the top three reported bottlenecks are:

The first one is an excuse; the real issue here is that the big data project is not led from the commercial business but from IT. When IT leads your data project then they can only aim for perfection, but there is no such thing as perfect data. You need to lead from the business in order to manage and understand the inevitable trade-offs in providing data and to be able to decide what is good enough. The outcome you want is not perfect data but profitable business change. Keep the end in mind.

The first two issues, then, are both about business leadership, or rather, the lack of it. That would be in line with our experience as well. We have found that successful big data implementations do five things right:

[Successful big data implementations do five things right]
  1. Lead from the top
  2. Deliver quick commercial wins
  3. Build the right IT capabilities
  4. Keep it business focused
  5. Engage the right skills

Lead from the top: Big Data opportunities cut across the enterprise and only the CEO and board can provide the leadership. Culture change quickly becomes important, and culture has to be lived from the top. Measure the right things: What is your organization’s return on data?

Deliver quick commercial wins: Every organization has low-hanging fruit; there is money being left on the table right now. Get started, now. The best time to start a big data project is ten years ago. The second-best time is today.

Build the right IT capabilities: You need the right capabilities to keep your commercial insights / data science teams productive in the longer run; this is where your capital investment goes. But build the right capabilities; these will not be the same for every organization.

Keep it business focused: Start with the end in mind, and the end is always business change. Not technology, data, models, reports, or anything else. Measure success by commercial KPIs like revenues and profits generated. Keep it real.

Engage the right skills: The key skill you need to successfully exploit big data is what McKinsey calls business translators who combine data savvy with industry and functional expertise. They bring it together and bring it to life. Don’t let IT lead your initiatives; don’t let Data Science run it, nor operations. This is a new function.

If data is an asset, then how do you measure your return on data? If you answer is around the size of your data lake, the number of data scientists you have, or even the number of cool-pilot-projects-that-never-got-into-production, then you’ve got it backward.

Measure your return on data with metrics that your CEO care about. Speak her language. Profits. Revenues and costs. Customers. New products, especially those that through innovation gives you a sustained competitive advantage which is hard for others to replicate. Make data a board-level issue.

If you measure your big data projects on anything else then you have already lost the game. You will not get leadership from the top, because you are not engaging the top. You will not deliver quick commercial wins, because you are so focused on your shiny new data technology that you forget about the business change, which is where the real work lies. You will probably over-invest in the wrong IT capabilities. You are obviously not keeping it business focused and you will focus on technical skills rather than the translator skills that are critical to make data successful in the business.

You will fail on all five points, unless you measure your return on data with metrics for your CEO.

You will fail to deliver business value and fail to return above your cost of capital. (Like 70% of companies in the study but that should not be a comfort.)

You will fail to deliver the innovation and growth that your organization needs to survive.

You can choose success. Winning with big data is not a game of chance. There are proven business models and battle-hardened templates and methods. If you do it right, your rewards will be immense. Back to the study and remember that even in 2014 leading telecommunications companies reported 20% of profits from big data. This is not some new industry, but one that has been around for decades. In the 3 or so years since the study, the technology has made giant leaps forward and the costs have gone down. What could you achieve today?

CYBAEA help organizations make money from data. Contact us and let's inspire you to success.