Organizations still struggle with making analytics projects successful. This blog explores how a survey gives us a clear indication of what are the top risks to account for. 

BARC publishes a BI survey every year. I was navigating the most recent BI Survey (2018), and I felt compelled to share some of the very relevant, empirically supported assertions it makes. 2,392 people answered this survey.

What are the top 4 most severe problems on Analytics projects?

1. Lack of resources on the project team (16%) 2. Unclear requirements (15%) 3. Software issues (15%) 4. Tight deadlines (13%)

I have managed many projects in this space, and this is right on. The right people will make all the difference. Having the right people will make sure you eliminate all or deal effectively with other problems. For instance, if requirements were not clear from the start, which competent team member would stand for that? And please stop imposing deadlines unless your company has made the transition to becoming an agile enterprise.

What is really interesting is that BARC looked at all companies but then also looked at the answers that came from those who have more success, or are "best-in-class." These were the answers:

1. No significant problems (63%!!) 2. Training-related issues (16%) 3. Lack of resources on the project team (15%) 4. Tight deadline (11%)

There are wonderful lessons to be taken from this which we probably all intuitively knew, like "competence matters". Experience seems to show that some of the risks materialize even for the most competent though. And, as expected, highly skilled teams do not seem to get caught with the basics, like unclear requirements or technology issues. To increase their chances of success, they deal with these risks proactively, up front.

So it becomes clear which risks your TOP four should be on any Analytics project. Let's look at them.

Number 1 - Lack of resources on the project team (basic) This one is in both lists, meaning that despite doing your best, it might still be an issue. As a VP of Consulting, it warms my heart that this is at the top. But seriously, and again you probably know this, but skills and availability are fundamental. If the project is worth doing, it is worth doing it right from the start. Just a few outside folks can make a huge difference to complement your internal team. It is people that will ultimately deliver your project. As my kids say to me often, "Remember that, now." Making sure you have the right skills and experience on your team goes as much for architecture and design expertise as for technical expertise.

Number 2 - Unclear requirements (basic) Asking to buy Tableau is not a requirement. So more than just documenting the requirements, doing it well is just as important. Agile methodologies help us in that they allow for the business to experiment and see what they ask for will look like, and then refine it. Delivering your data analytics projects in agile fashion is probably your best mitigation for this risk.

Number 3 - Tight timeline Put it on the list. You will not necessarily convince executives to change their minds if they impose some artificial deadline on the project without some issue as support, but seeing it on the list will sensitize them. Again, an agile company culture (not just as a method on the project) will go a long way to mitigate this risk as well.

Note: if you've noticed, you can address two of the top three risks by adopting an agile method. But it introduces a new danger - the clash of cultures if the company is not ready for it. I can't emphasize this point enough.

Number 4 - Training Training is probably not one you expected to see in the top 4. But it makes sense. More and more, projects are expected to generate a planned ROI, and your finance department will come and verify whether the project did achieve it. And if the new technology is not used as intended, with the right mindset, and by the people you expected would use it, then that ROI will be harder to reach. Have a solid Training and Communication plan. Define your user personas for your requirements and then add a section on how you will communicate with them, train them, and entice them to adopt the new solutions. From catalogs and "data marketplaces" to self-service analytics and sandboxes, these can all seem quite daunting for even skilled employees. Think of this part as follow-through, making sure that what you deliver is going to have a REAL IMPACT.

Conclusion An initial risk management strategy should identify 5-10 risks, no more. These should be managed actively. There is not much point in making an extensive list and then never looking at it again. And when it is for an Analytics project, now you know, based on real research, which risks are your top four.

©2020 by Modern Data Analytics