According to the Oxford Global Projects Database, 99.5% of large projects fail. According to Harvard Business Review:

Consider the data [from the Oxford Global Projects] database that includes more than 16,000 projects—everything from large buildings to tunnels, bridges, dams, power stations, mines, rockets, railroads, highways, oil and gas facilities, solar and wind farms, information technology systems, and even the Olympic Games. Collectively, it paints a portrait of big projects across the world. And the portrait is not pretty: Only 8.5% of them were delivered on time and on budget, while a nearly invisible 0.5% of projects were completed on time and on budget and produced the expected benefits. To put that more bluntly, 99.5% of large projects failed to deliver as promised.

Forbes reports slightly different statistics, but a similarly bleak outlook for IT Projects. Some projects can go so wrong as to threaten a company’s existence. KMart’s massive $1.2B failed IT modernization project, for instance, was a big contributor to its bankruptcy.

These statistics should give any executive considering significant IT investments pause. Businesses should expect their data analytics projects to fall short unless they take extraordinary action to ensure success. Despite an alarmingly high failure rate, new-tech hype seems to outweigh the discussion of successful IT investment strategies. We’ve become so accustomed to IT projects not meeting expectations that we take missteps for granted. In a recent conversation, a VP of Finance dismissed his concerns about the progress of an IT  investment worth hundreds of millions of dollars with a shrug, as if to say ‘What can I do.’

I want to help. In this article, I’ll share insights to improve success rates gleaned from over a decade of managing and working on Data Analytics projects.

Why Data Analytic Projects Fail

Data Analytics, including Data Science and Artificial Intelligence (AI), represents one of the most significant opportunities and yet one of the biggest risks in business today. Data Analytics projects come in late, are over budget, or fail to fulfill their purpose for a few reasons.

Big Investments Require Careful Oversight

First, giant consulting firms like PWC or Accenture usually implement big IT projects without proper oversight. Keep a close eye on these firms. They’re motivated by their own goals and incentives. Going over budget and doing unnecessary work make them more money.

You need folks without a conflict of interest to oversee these projects. That could be internal employees, or you can hire a boutique consulting firm. I wouldn’t recommend hiring another large firm to oversee the project since that could lead to infighting. But, small boutique consultancies know they don’t have the size to take on large projects, which reduces infighting.

If you want the oversight completely internal, make sure your employees are up for the challenge. Don’t appoint someone just because you like him or because he’s been working with the company for twenty years. You need overseers with strong process engineering, financial, and technical skills. That requires executives to be honest with themselves about whether they know enough about Data Analytics to staff and manage teams able to overcome a 99.5% failure rate. Most executives don’t because that’s not their job.

Here are other common reasons why Data Analytics projects don’t meet expectations.

Functional Disconnect

Data Analytics requires a deep understanding of the data and processes, which typically reside within a given function, department, or in the heads of a handful of subject matter experts (SMEs). Data Analytics Projects often fail due to a failure to communicate.

For example, for years, a major bank asked their salespeople to review contracts by hand and type data into a web front end for regulatory reporting purposes even though the raw data lived in another system. I successfully led a complex system integration project to get the data. The technical piece was easy. Translating the regulatory lingo in light of the bank’s loans and managing the people aspect of getting four different departments to sign onto the integration was hard.  I had to spend months analyzing data, working with SMEs to understand what was happening on various systems’ back end, and giving presentations to technical leaders, getting them comfortable with integration logic that could be used in a court of law. Bankers are correctly very cautious and careful about not getting into trouble with regulatory authorities. Lots of communication, patience,  and a thorough and transparent testing regimen finally got everyone on board and allowed us to sunset the manual data collection process.

Another project building out an online store was delayed due to miscommunication between the front and back end systems. The online store needed to pass customer orders along to the back-end fulfillment system. The project was organized so that each project manager and the technical team worked on their own scope and weren’t talking. So, the integration of the front and back-end systems wasn’t explored thoroughly, and the technical teams had to redo a lot of work to make the systems talk to each other.

Financial Implications Take a Back Seat

Data Analytics projects are somewhat unique in that they’re almost always initiated with the expectation of clear financial benefit. The intent could be to give the salespeople more information so they can sell the company’s goods or services at a higher price. Or perhaps the system will provide information to enhance the effectiveness of ad spending so the company can sell more without increasing ad spend, etc.

Despite clear financial motivation, I rarely see Financial Planning and Analysis (FP&A) folks on Data Analytics projects. They’re almost always implemented by people with computer science backgrounds. I’ve always found it odd that companies invest hundreds of millions of dollars in IT projects with little FP&A oversight. Now, it’s a fact of life that any large project or initiative requires compromise and adjustments. And usually, those compromises or adjustments are made to make the technical team’s life easier. Unfortunately, computer scientists have little insight or interest in the functioning or profitability of the business. So the adjustments they make often have negative consequences on the project’s ability to increase net income. For example that could manifest itself in a lack of data quality, limitations in how the analytics will plug into a business process, or some limitations in ease of use.

Most executives have seen an incredible Data Science presentation or a million-dollar dashboard that no one uses. Successful data analytics tools drive bottom-line results; anything else represents money wasted and a missed opportunity. How data analytics becomes actionable depends heavily on the audience, industry, function, etc so it’s hard to generalize that part. But, it’s a waste if it isn’t used

Thinks Slow, Act Fast: How to Ensure  Successful Data Analytics Projects

To ensure on-time, on-budget, and effective Data Analytics systems teams need strong leaders who understand the technical and business aspects of Data Analytics projects.

The lack of financial value add is still a big gap today. Most executives can point to hundreds of millions of dollars worth of disappointing IT investments, as evidenced by the 99.5% failure rate documented by the Oxford Global Projects Database. Yet, I still scratch my head and wonder what executives expect from teams led by computer scientists with little business acument or FP&A support. Careful financial planning and analysis are needed to understand how data analytics investments add to the bottom line. And once the plans are in place, finance teams should measure and report on the project. For example, do the salespeople use that million-dollar dashboard to upsell customers, or do they see it as a nuisance and waste of time? Is that expensive digital marketing initiative working? Is the ROI on your ad spend actually improving? Does the finance team even measure ROI on marketing campaigns (probably not)?

Project delays most often result because some challenges arose that took more time than anticipated to address. Agile project management methodologies have largely come about to sidestep the complexity and fast pace of change in today’s technology space. An alternative to Agile and a way to manage complex projects to a specific deadline would be meticulous project planning by technical experts and prototypes exploring any anticipated challenges or major new technologies. But that would cost almost as much as the project itself. So, I think a well-run Agile project, and its resultant loose deadline, is probably the best we can do.

Success Factors

Executives should ensure the success of large Data Analytics investments by appointing leaders skilled in both technology and business, particularly strong FP&A skills. They also need strong process engineering and project management skill sets. FP&A teams highly versed in the relevant technology should evaluate and actively measure these huge investments’ success.

What do you think?

Thanks for reading to the end! This blog is my project in the pursuit of truth. I spend dozens of hours researching each blog post, so I hope you found something useful.

Our click-bait culture needs good ideas in an increasingly complex world. That depends on good men and women engaging in intellectually honest discussions, sharing ideas, and challenging each other’s thinking. Writing out my thoughts in detail, along with lots of research, helps me arrive at a more accurate view of truth based on well-documented facts.

If you’d like to support my blog please buy my book. Here’s the link!