The Elephant in the Data Analytics Room

Most data analytics projects fail. Only 1 in 5 projects will deliver any meaningful business outcomes. Data Science, a newer darling of the analytics space, fares even worse; Per MIT Sloan, only 11% of businesses will report any financial benefits of AI, and Gartner predicts that 85% of such projects will actually deliver erroneous outcomes.

Despite a ton of failure and cynicism, Business Intelligence investment is as high as ever. In fact, even in 2020, with the uncertainty of the COVID-19 pandemic and an ongoing distrust over current data projects, most businesses report that they will actually continue to invest more in data analytics and machine learning.

So, why this seemingly irrational behavior? Why do business leaders consistently make what appears to be one of the least data-driven decisions possible—the decision to keep investing in data and analytics itself?

The truth is that there simply isn’t an alternative: To answer the most basic questions about your business—will we hit our numbers, what types of demand gen efforts are most valuable / effective, who are our top performing teams / channels / segments etc — you must have and must make use of data.

Organizations also know analytics investments provide non-linear returns, an exponential progression often drawn as a maturity curve. When businesses can find a way to create a competitive advantage using data, ROI is through the roof. In the world of customer analytics, McKinsey came to the conclusion that businesses that are effective in using data are 23 times more likely to acquire customers and nine times more likely to retain them.

But the reality is that most businesses never make it to this level of maturity. Despite a ton of investment and innovation, businesses of all sizes continue to find their data engineers and analysts forever stuck on the most basic of reporting requests—financials, sales pipeline, opportunity win rate etc.  


We all think it, so we might as well say it: analytics is hard, expensive, and we will never really be “done” implementing it.

There are more data sources now than ever, each of which continuously evolves and requires ongoing maintenance. Data itself is more complicated and often has to be shaped for analysis. Translating business questions into data analytics requires a combination of technical capabilities and domain expertise. And then there is the semantics of business metrics itself, and it is an all-too-common scenario for business leaders to be loose with their definitions of what the numbers mean. 

Today, each of these tasks represent additional tools, an additional layer in an analytics stack that you must implement and integrate to deliver value. Given this, it really shouldn’t be a surprise that failure rates in data analytics are so astronomically high.

This is the real problem: the focus on analytics tooling itself. We are measuring the progress of data analytics by the deployment of technology rather than by the business outcomes we want to achieve.  

I believe there are three keys to tackling this problem:

  1. Work backwards. Don’t start your analytics journey looking for a Data Warehouse, a BI platform or an AutoML tool. Start with the business questions you need to answer, and who needs to answer them (e.g. Will we hit our numbers, which customers are likely to churn, what can we do about it).
  2. Recognize that many of your top questions are not unique to your business. There is no competitive advantage to be gained from you reinventing the wheel. Believe it or not, every sales organization needs to figure out whether they have enough pipeline and forecast whether they’ll hit their numbers with a combination of top and bottom-down analysis.  Every marketing organization needs to figure out what the most effective campaigns are and track the efficiencies of their campaign. Look for an analytics solution that embeds the best practices regarding how to answer these universal questions, so that your analytics investment can go towards truly value-creating efforts.  However, make sure that the solution you select is customizable, open and extensible so that it can scale and evolve with the needs of your business.
  3. Identify the data questions that are truly unique to you. These are the questions that will net your business a true advantage over the competition. Task your analysts and data scientists with solving those questions, and only then, focus on the analytics tooling they need to be successful.

For all its frustrations and failures, I believe we’re still at an exciting time in data analytics. In my 15+ year career building BI tools, I’ve learned lots from my conversations with hundreds of thoughtful and data-driven leaders out there. I’m inspired by the increasing value they place on data literacy through their organizations and the creative ways they use data to gain a competitive edge.

Sales leaders want to moneyball their sales team. Marketing leaders want to drive opportunities. Revenue leaders want to optimize their funnel. And that desire continues to motivate a growing market of analytics tools to help leaders solve those problems.

More than ever, every single one of us wants to make better data-driven decisions and challenge our colleagues to do the same. Let’s start by focusing less on the analytical tools, and focus more on questions those tools should help us answer.

Learn how Datajoy can help you!