How to not fail when launching data analytics programs.

In life we spend a lot of time thinking about how to succeed and that’s usually a great starting point in mapping out a plan.

However, there are cases where the best place to start is by thinking about how to avoid failure. This is almost always true in cases where the cost and likelihood of failure is high.

After 15 years of developing and implementing analytics programs I believe they fall into this category of “avoid failure” and the evidence seems to support my hypothesis.

Here’s a quick visual to drive home the point.

analytical program success failure 2.gif

Failure is currently the norm but why is this with so much talent and money going into analytics right now?

Most will tell you it is complicated (it is) but I’m going to share an approach to avoiding failure in implementing analytics programs that is actually quite simple (not easy) and relatively affordable (but not cheap).

Failure Facts

Late in 2017, Gartner estimated that big data and analytics projects fail to materialize 60% of the time. That figure was later updated to be 85%!

Gartner’s top data and analytics predictions for 2019 include “80% of AI projects will remain alchemy, run by wizards whose talents will not scale in the organization through 2020.”

According to the 2019 New Vantage Survey, 77% percent of respondents to their annual survey, say that business adoption of analytics continues to be their biggest challenge.

Even more troubling, the results from a 2018 Chief Marketing Officer Surveyconducted by Duke University, Deloitte LLP and the American Marketing Association found that less than 2% of marketing leaders believe their companies have the right talent to leverage marketing analytics.

How To Fail

There are many ways to fail and there are many reasons why the failure rate is so high! Here’s the all time greatest hits for how to fail in analytics.

Let IT get in the way. Technology solutions looking for a problem is one of the most common issues associated with analytics failures. Designing and implementing a system for business intelligence and data analytics before you know what you need and want is a sure fire way to fail. I’ve seen some massive failures where millions of dollars and countless hours were wasted because of this issue. Most anyone working in this field could likely say the same. Don’t lead with IT.

Work with whoever is currently around. Highly capable analytical talent with industry specific expertise is rare, difficult to locate and not likely just hanging around. The good ones are in high demand so you can expect to pay for it. This and other real factors mean most projects are staffed by current team members. While your people may be great at what they currently do (which, they will presumably have to continue doing), and even better if they are good analysts, not taking seriously the independent discipline and the level of effort associated with implementing an analytics program is a fatal mistake.

Photo by  Brandi Ibrao  on  Unsplash

Believe you know it all. Data analytics is a complex space with regular technological and methodological advances that are too much for most teams to work out for themselves. Some examples might include how you could use text mining and natural language processing to better understand your customers or the market or utilizing new products like Google Data Studio to create easily accessible dashboards in a jiffy. The options are seemingly endless. More likely than not you have a dearth of analytics knowledge in your organization so before you think you’ve got it all figured out, think again.

Don’t account for flexibility. What you need today may not be what you will need tomorrow, literally. The ability to add and remove capabilities and capacity when needed is fundamental to making the most of your analytics investment. One example I’ve seen time and again is companies hire an analyst in hopes they have the right skills only to find out that that their limitations become your limitations which often leads to floundering initiatives.

Don’t get the right kind of help. Jumpstarting these initiatives in-house proves impossible for many. Traditional consulting arrangements require lengthy contracts and high levels of effort to manage. On top of this, many firms are working harder at keeping their team on your payroll even after the work has been completed. This can lead to protracted engagements that find you doing more for the consultants than they are doing for you.

How To Succeed

While success doesn’t come easy, this approach is directly related to the reasons for failure and is based on 15 years working from within with large and small organizations and as a management consultant.

Start using [your] data today. The journey of 1,000 miles begins with the first step. Data analytics is no different. Begin by taking stock of what data you have and begin working with it using whatever tools you and your team are familiar with. Don’t worry about its limitations or yours. Get excited and get moving!

Get curious. Now is the time to start asking questions, reading (looks like you can check that box right now), engaging with experts on social media and branching out. Start asking the really important questions about your customers, the market, trends and opportunities for the future. These questions are the starting point for shaping your analytical needs, creating your strategy and maximizing the impact of your investment.

Don’t let IT get in the way. Resist the temptation to make IT the focus. Do not fall victim to regular sales calls from vendors trying to sell you the latest BI platform. That time is spent far better working with your data, your team, and learning more about what is needed and possible.

Use lightweight tools. This is the flipside of not letting IT get in the way. The world of analytical tools is becoming richer by the day. Automated machine learning platforms are speeding the time from data collection, cleaning and loading to insight like never before. Using basic tools and techniques for each stage of the analytics process (question formation, data collection, preparation, analysis and visualization, repeat) is your best bet. Any tools that are not working for you can be easily discarded for something newer, better, lighter, faster or easier. Don’t be afraid of open source software. Some of our favorite tools include the KNIME analytics platform, Tableau and Microsoft’s Power BI. That said, nothing beats the power and flexibility of coding in R or Python, if you have the skills.

Get the right help. Even with the advent of powerful and lightweight tools, one of your primary needs will be in finding the right partner to help support you in this journey. The partner you look for should be able to dedicate a team to your effort including an analytics translator, data scientist, data analyst and visualization specialist. They should support you in all phases of the data analytics lifecycle including doing the analytical work. Your contract with them should also include an ad-hoc support option for brainstorming and responding to questions as they arise. For many companies, outsourcing analytics is a great options for reducing risk and quickly getting insights you can use. Our firm, Primary Notes, specializes in this style of support.

Visualize and communicate. Great data visualizations amplify great insights. Data storytelling and visualization are fundamental to the success of your analytics program and special emphasis should be placed on this from the start. No matter how good your analytics are, if they do not manifest themselves in compelling, well designed and presented visualizations, your ability to drive action from the insights will be hampered.

Maintain a business first focus. Your team should always be lead by questions that are important for your mission, not in using technology and analytics for their own sake or simply because they are popular. A business first focus will ensure you are articulating problems that need solutions rather than the opposite. Make it a regular part of your leadership meetings to capture any questions that come up whether or not you think they can be answered by data analytics. These questions should be available to everyone on the team and everyone should be able to provide their input as questions arise.

Iterate. Experiment. Fail. Succeed! Data analytics is a practice which means that you need to exercise these skills repeatedly and regularly in order to improve or maintain proficiency. Regular experimentation with the types of questions you ask and the analytics you perform will keep things interesting for everyone, develop new skills and insights and bring you closer to meeting your ultimate objectives (in addition to avoiding the dreaded failure).

The most successful teams will deliberately stretch themselves so when failures occur they see them as high learning potential that will facilitate growth. Leadership should support this high growth mentality and be invested in times of success and failure.

Don’t Be Another Statistic

Success in data analytics, like in business and life, is hard fought and requires work and focus. Remember, there are no [IT] shortcuts but with clear direction, knowing what to avoid and what matters and getting the right help you’ll be on your way to avoiding the dreaded analytics failure that is so common.