Why Couldn’t My Team Do That?!
Delivering transformative analytics isn’t about being better. It’s about doing things differently.
Over the past year, our data science team has delivered nearly 500 analytic use cases to a variety of private and public sector organizations, including some of the largest, most successful companies in the world. We have even found ourselves tackling problems their internal teams have already tried unsuccessfully to solve. The depth and breadth of this experience has given us a clear understanding of what it takes to successfully deliver use cases — and it may not be what you expect!
Our typical engagement begins with an intense, 1 to 2-month sprint where we develop a minimum viable prototype to prove the feasibility of the solution. The culmination of the effort is often a meeting with the project sponsor and other senior executive leaders, during which we share our insights and the business impact we believe we can drive by scaling the prototype. In one of these recent meetings, a senior client wondered aloud, “Why the h@$% couldn’t my team do this?” This is a common question we hear, sometimes from the leader of the data science team, but more often from the CFO, COO, or even the CEO.
Five Ways to Empower a Team
Senior executives are typically familiar with the investments made in building the company’s data lake, acquiring the right tools, and hiring the right internal team. When a project fails to deliver impact, leadership may be tempted to question the capability of their internal team. Our experience, however, tells us that the problem is rarely at the team level: The data scientists we encounter are experienced professionals with deep technical expertise. Why then is our small team able to succeed where an entire internal team is challenged? We’ve come to identify a five common factors:
1. Focus. Our team is 100% focused on solving a specific business problem. Internal teams are typically faced with competing projects and priorities that draw their attention in multiple directions. Our data scientists often hear from their client counterparts how much they wish they could dig deeply into one problem for several uninterrupted months. The lack of singular focus is particularly problematic when it comes to the long-term, strategic issues our teams often tackle. Although addressing these larger issues can have the greatest business impact, they are often down-prioritized in the chase to put out short-term “fires.” These constantly changing priorities and lack of contiguous focus can make meaningful progress difficult.
2. Driving business outcomes. Typically, our team is engaged to solve a very specific business question. In our experience, internal teams are rarely presented with such discrete objectives. While they wait to be given clear business problem to solve, business units wait to be given ideas and suggestions for problems that can be solved. This creates an unintentional stalemate that leads to internal teams being asked broad, open-ended questions related to making use of data (e.g., “Now that our data is in the cloud, what can you do with it?”) or they are asked to automate reports and analyses. Internal team members may find themselves trying to find the right problem to solve, which may not be well-aligned with senior executive priorities. As a result, the impact of their work may never be fully realized. Worse yet, their work may prove to be completely irrelevant to business goals.
3. Senior attention. Our project sponsor is typically a senior executive, virtually guaranteeing visibility among and access to the senior executive team. This inherently means that any blockers (such as data access or process-owners time) or resistance among leaders or staff can be addressed immediately. More importantly, it gives us a mandate for change. We have the license to change underlying business processes and decision making to realize the potential of the analytic solutions we create. Internal teams rarely get this level of senior attention. Their progress may be slowed for weeks or even months as they try to navigate issues without the proper mandate. Further, many times these teams are 3–4 layers removed from senior executives. Attempts on their part to engage at the executive level may require a drawn-out series of “escalations” to penetrate numerous organizational layers.
4. Change management. We have consistently seen that the effort required to implement an analytics use case breaks down to 10% algorithms, 20% data and technology, and 70% change management. Many organizations fail to plan for this degree of change management and, as a result, fail to realize the full potential of the analytic solutions that are created. Furthermore, the existence of organizational silos typically means that the internal analytics team is limited to answering the analytic question — just 30% of the effort. They are not empowered to deliver the other 70%, the part that is critical to effectively deploy solutions. Instead, they must convince other organizational elements to prioritize the effort and make the required changes. Our role transcends organizational silos and hierarchy, allowing us to implement the technical solutions while tirelessly driving the change-management journey to shift how our clients do business.
5. Hybrid, fully integrated teams. Because of our position as outsiders, we can side-step typical organizational silos and blend data scientists, data engineers, software developers, and MBA business consultants into a single, hybrid delivery team. This prevents the friction points and process delays we often see in client organizations when solutions are handed off from data science to IT to business process owners. Our team’s blend of expertise also encourages a creative tension that unlocks significant value: MBAs will not sleep unless they see how analyses are relevant to driving business value, and data scientists won’t be satisfied until they have the right technical answer. We also deploy our teams into the field, where they can work hand-in-hand with business process owners. This integration gives us a deep understanding of the nuances of the business problem and surrounding issues, thus ensuring that our solutions are practical and address barriers to implementation. Most internal teams, on the other hand, are co-located in a central location without the motivation or budget to travel into the field, further preventing this level of integration and joint problem solving.
Every organization is different, but these five factors can help explain why so many senior clients wonder why their internal teams aren’t as successful as ours. The good news is that it is possible to address all of these factors in just about any company. Our most impactful efforts are when we work with our clients to help them break through and build these practices into their own organizations. We’ve helped many clients strengthen their internal data science capabilities and build effective cross-functional teaming — and they don’t need a long-term change project to deliver near-term results. In our experience, simple steps related to governance, prioritization and sponsorship consistently yield benefits. The scale of these changes is far less important that ensuring that they are implemented with a constant focus on business outcomes. Many organizations have made significant investments in building data science teams and arming them with the right set of tools and technologies. Now they need to take a few organizational steps to unlock the true potential of their internal teams and enable them to consistently deliver the same level of impact our project teams routinely deliver.
As organizations begin to shift their focus towards broader transformation to drive enterprise-wide impact, a new set of challenges will emerge. In a recent article jointly written with MIT, our team highlighted what we consistently see successful companies doing to scale AI across their organizational. Whether focusing on individual projects or enterprise-wide transformation, no challenges is insurmountable if leaders are thoughtful and intentional in their approach, and consider organizational as well as technical barriers to success.