Our recent white paper, Taking the Lead with Analytics, shows that those delivering sustainable business value from data and analytics (D&A)– who we’ve defined as “leaders”—commonly exhibit trust in their actions, from data collection to data-driven decision making. Within that piece, we offer a deeper look at the beliefs and behaviors that separate leaders from laggards in establishing trust. Over the next four weeks, we’ll be bringing you a series of collaborations with experts across North Highland to offer deeper insight into the steps that organizations can take to build the trust that’s so critical to D&A leadership. To start, we’ll look at the role that trust plays in organizational structure.
Centralized and Closely Aligned
In designing a D&A organization, the question can often become overly simplistic: should analysts be centralized in a single function or embedded in supporting roles to business teams? We reject this as a false dichotomy and answer “both.” Both survey data and case studies back up our stance.
Centralizing the D&A team is an important step for both cultural and technical reasons. Companies that treat D&A as a supporting role reporting to business leadership struggle to consistently deliver a return on investment: in our survey, a company reporting that D&A is a centralized enterprise function is 3.77 times as likely to be a leader than a laggard. In a podcast interview by Kathryn Hume of integrate.ai, Eric Colson, the Chief Algorithms Officer for Stitch Fix, explains that their “algorithms team… [is] its own department, it’s not buried under engineering or finance or marketing. It’s a standalone thing and equal status with all the other functions.” In addition to establishing D&A as a centralized, standalone team, it’s fundamentally important to instill a culture where D&A and the business are closely integrated, and where D&A has equal weight and influence in strategic decision-making.
On the other hand, analysts must also be embedded or closely aligned to their business partners. Failing to meet internal customers where they’re at degrades trust and can quickly relegate D&A to a rigid, ticket-taking function, usually tucked away in IT and engaged via cumbersome requirements and estimation procedures. In our survey, when asked “Which of the following do you believe are true for your analytics employee(s)?” 53 percent of leaders (compared with only 30 percent of laggards) ranked “They are empowered to meet the needs of their internal customers” in their top three responses, and fully 24 percent of leaders ranked this as their top response.
The best model blends both approaches. One needn’t look too far to see this model in action, including at Stitch Fix. “We’ve structured [the] algorithms team such that there is alignment to the various other departments… This way there [are] relationships at play [such] that people get used to the domain and to the various players and they can work closely together… it doesn’t feel as much like projects but an ongoing initiative,” Colson says. “They are still a central team, but they are aligned with various partners in the company.” Riley Newman at Airbnb similarly explains their “hybrid centralized/embedded structure: we still follow the centralized model, but we have broken this into sub-teams that partner more directly with engineers, designers, product managers, marketers, and others.”
Effective structures blend the elements of a centralized and embedded model. The precise structure will vary between organizations, but the imperative for delivering trust across the value chain provides organizational leaders with a common set of principles to follow:
- Centralization is often a key first step in developing D&A maturity, and is important to maintain for attraction, retention, developing the team (e.g., career pathing and skills development), and developing standards for how work gets done.
- Business alignment is key for establishing trusted partnerships and ensuring D&A work hews closely to top business priorities and thereby drives value. As we’ll explore in the next part of the series, this often includes the involvement of visionary business leads responsible for pulling D&A business cases into the organization.
Organizing for Mastery
A related consideration is the D&A organization’s operating model. Our white paper makes a clear case for using Agile methods, and these are enabled when analysts are tightly connected with their business partners, not taking requirements “thrown over the fence.” Says Colson, “We really value iteration… [Specialization] works for assembly lines precisely because all the requirements are known up front… Data products are different – we usually don’t know exactly what we want. We need to iterate to find… the real requirements. Iteration requires fluidity across the different functions…”
While Agile is a good first step, embracing DevOps methodologies can generate even greater value. DevOps incorporates and extends many Agile principles. A key element is the use of generalists who take responsibility for their solutions through deployment, as Colson explains: “The other unintuitive part is not only are they faster, but they’re better. They’re better because you now have this technology stack that’s pieced together by one person’s vision… it gives the developer holistic context and encourages elegance between parts of the tech stack…” On the same question about their employees as above, “They continually seek to optimize the system” was a top-three response for 59 percent of leaders and only 30 percent of laggards, speaking to greater autonomy and broader responsibility across the analytics value chain. Indeed, 74 percent of leaders and only 29 percent of laggards report that they instill autonomy in employees “often” or “always” in the delivery of analytics initiatives. “We aspire to have somebody who loves the research but also loves the impact… Nobody else wants something handed to them – it seems foreign… they’re not aware of all the context or the passion that went into the framing,” says Colson. “We want high job satisfaction, and usually that comes with ownership from end-to-end.”
The importance of organizing to deliver D&A is clear: Forrester reports that insights-driven firms are growing at an average of more than 30 percent annually—eight times faster than GDP. Any organization seeking to establish leadership in D&A by building trust across the analytics value chain should work to implement the principles we’ve outlined. So what do these principles mean for business partners who aren’t responsible for the D&A team’s organizational structure? In the next installment, we’ll explore key success factors for the D&A-business relationship through the lens of decades of IT transformation successes and failures.
This post was co-authored by: Dan Montgomery
Dan is North Highland’s Data & Analytics Information Management Lead. He brings more than 25 years of experience in the management and implementation of business-driven information technology solutions. He is a highly engaging leader and consultant serving Fortune 500 companies and public sector clients with expertise in business intelligence, data warehousing, business and technology strategy, merger and acquisition, system integration, information security, and project management.