Data analytics projects are becoming increasingly prevalent as companies realize that tapping into data can revolutionize the way they make business decisions. As tools become more accessible and projects develop, it’s important to pause and take stock in what makes a project a success or a failure; one should never rush into a project without proper planning and support.

At BDO Digital, we’ve worked on a lot of projects over the past 35 years. We're proud of our wins, but we can also say we've learned a lot from some missteps along the way. Taking a page from Star Wars, Game of Thrones, and The Walking Dead, here are three examples of how poor architecture can derail a project before it even gets off the ground.
 

Problem #1: Poor Beginnings

The Empire wanted to create the ultimate weapon to continue their quest of ruling the galaxy. They built the Death Star, but their impatience for galactic dominance led them to jump right in without a solid understanding of the current environment, market, customer needs, and competition.

As a result, there were oversights. During the Battle of Yavin, Rebel Alliance competitors exploited the Death Star’s weakness, firing a proton torpedo into a thermal exhaust port and destroyed its main reactor, resulting in the Empire’s Project Death Star Failure.

Who is to blame for this project failure? All fingers point to the Architects… I mean, who left that thermal exhaust port vulnerable in the first place? And not once, but twice (spoiler alert: Starkiller Base had a similar problem with a vulnerable oscillator 30 years later)! I mean, hello?? I want that Architect Captain in my office RIGHT NOW.

To be fair, some may argue the fault can’t solely be placed on the architect. After all, if we learned anything from Rogue One, it's that there are always multiple factors at play. For example, the so-called altruistic Galen Erso intentionally sabotaged the architectural design to leave room for future failure. So you could argue that a lack of leadership oversight and security were also a major problem. But for the sake of this analogy, let's take a look at what the architect could have done differently.

Lesson Learned: Start with the Right Approach!

• Seek to Understand your Users, Customers & Competitors

Too often as architects we feel pressure to design and build solutions for the 1% and not address the needs of the 99%. We frequently witness architectures that were destined to fail due to complexity or too much emphasis on features that most of the users don’t care about. For example, some users will say they need real-time dashboards and will accept no compromises. The architecture required for real-time vs. hourly refreshes is substantially different, both in cost of services and resource time to build such as solution. Make sure every architectural decision is backed up with honest ROI analysis. Ask reasonable business questions that can expose the financial value of the features.

• Ask the right questions to the right people

Make sure you are talking to the users who will use this data to make business decisions. IT is just as susceptible to group-think as anyone else and we see much more success on analytics projects when the users impacted by the solution are brought to the table during discovery and requirements gathering.

• Pragmatism and understanding help tailor a customized approach

I can’t stress the pragmatic approach enough. Sometimes it feels like everything will be easier if we just get this first pilot into production, but the journey before that launch can daunting and exhausting, especially if this is a first-time launch of an analytics project. Be prepared to make compromises and anything that gets in the way of success should only dealt with if success is going to be a stake. Everything else goes right to the product backlog.

Examples of Failed vs. Successful Architecture

• Fail – The client ran into data sampling issues with the Google Analytics API.
• Success – Consume Google AdWords API on a recurring incremental schedule and store history in a client-owned data store instead of expecting the Google Analytics connector to be the source of truth.

 

Problem #2: Lack of Unity

Competing priorities can be another wrench in the gears of project development. In Game of Thrones, each House has their internal goals which often conflict with the goals of other Houses in the realm. The same is often true within a company – teams that should be working together for the greater good of the whole sometimes lose sight of the big picture when departmental needs and politics take precedence.

But if House Operations isn’t on the same page with House Accounting & Finance, and House HR is battling House Sales & Marketing, they won’t see the bigger threat that is project failure.

Gartner predicts that 60% of big data projects will fail in 2017. Projects cannot succeed without engaged shareholders across teams. Data is often sprawled and stored across multiple teams. It needs to be consolidated and to do this, you need everyone on board.

Lesson Learned: Unity is Key!

• Before beginning any project, departments need to communicate and have a unified, holistic approach
• Change Management is a best practice to facilitate communication, education, and end-user adoption and consumption
• Every project should include an OCM plan to ensure user enablement and adoption.
• Data should be unified in one platform, reports coming from a “single version of the truth”

If you want your data project to be one of the 40% that succeeds, you’ll need buy-in and unity from the organization as a whole.

Examples of Failed vs. Successful Architecture

• Fail – Every department had their own extract of sales information. Marketing used a view that IT built for them, Sale used a simplistic query that included some bad data (canceled orders) and Operations had an order query that excluded certain internal product orders (intercompany inventory transfers)
• Success – One data repository for all departments, one model for how to measures sales and orders, and one data integration pipeline for loading sales data.

 

Problem #3: Ever-Evolving Security Threats

The threat of cybercrime continues to rise as hackers adapt to new security measures and take advantage of changes in user’s online behavior. Increasing Ransomware attacks can cripple businesses, sometimes with dire consequences. Security has to be a priority when architecting your project. If not, you’re as good as you are when you're stuck in an unprotected barn with only worms to eat, surrounded by post-apocalyptic zombies shuffling forward to destroy your data project.

If your organization doesn’t have any internal or external security, well, you’re not long for this world.

Perhaps you will implement some security measures that will help you feel safer and finally breathe a sigh of relief. But external security issues still exist, such as phishing scams, malware, or psychopathic, one-eyed Governors trying to overthrow you and your team from your fortress. To be successful, you still have to deal with that.

Just when you think you have it all figured out, new threats emerge!
 

Lesson Learned: Smart Security Protects the Data We Love!

To ensure your data and people are safe, the proper security needs to be built into the architecture.

Data Security & Application Security

• Row, Column, and Cell-Level Security
• Power BI Apps and Content Packs

Security For/Against

• Privacy – Think Payroll or Sales Commissions
• Malice – Ex-Employees, Hackers, and Competitors
• Segmentation – Manage access for the right people

The Right Security Posture is Key

• Apply the right amount and not the most amount
• Never “Set and Forget” – Recurring Security Audits
• Apply separation of server admin and content admin

Examples of Failed vs. Successful Architecture

• Fail – The client tried to use the free version of Power BI to build their analytics. Data and company IP on business rules was easily lost to terminated employees, old customers, and possibly some vendors they no longer work with.
• Success – Use a professional version of Power BI such as Power BI Professional or Premium. Institute reasonable Power BI governance procedures and processes. Don’t give everyone copies of your data, give them governed connectivity and block extracts where appropriate.

If you’d like to discuss how your organization can benefit from advanced analytics and make sure your architecting projects to succeed – or if you just want to talk about fan theories, contact BDO Digital to discuss.