Webinar: Why logical layers matter, and how to use them -Watch now

3 Analytics Leaders on Building Efficient Teams

Image of author
Tayler Mehit, Associate Marketing Manager

July 18, 2018

5 minute read


What makes an analytics team great? To find out, we interviewed the leaders of three:

Our discussion ranged from team structure, to mission-driven motivation, to measurement of success. Their responses can help other folks steering analytics ships of their own form high-level philosophies to keep their teams on track.

Creating structure

All the data team leaders we interviewed agreed that successful teams are built with the entire organization's needs in mind. Church's team of four data scientists at Patreon, for example, needs to deliver on all internal analytics requests at the company. When asked how her team fits into and supports the rest of the organization, she said:

“The data science team at Patreon works as a horizontal layer, supporting all teams with the data they need to make great decisions. Most of our work focuses on providing the product team with data and analyses, whether it's at the beginning of a project in the research phase or towards the end, analyzing the results of a randomized experiment. We also spend some time supporting Patreon's operations functions, such as helping the marketing team understand how well our blog acquires creators or assisting our sales team in the creation of their data pipeline.”

Analysts who serve the entire organization, like those at Patreon, have the benefit of context. Analyst teams that support a wide range of functions get a birds-eye view of organization-wide analyses and a high-level view of the company's goals. This complete picture helps otherwise distinct teams, such as sales or product departments, to move in tandem. For example, Gangal's team at Fundbox worked across multiple teams such as Marketing, Sales, Product, Risk, and Servicing and Collections. All of the leaders we interviewed opt for an approach that covers the needs of the organization at large.

Kogel's team at VSCO is comprised of four data scientists. Though they work to support the entire organization, they have a particularly tight focus on Product, Business Operations, and Marketing. His team is armed with a working knowledge of the organization's data and a direct line of communication between stakeholders. When describing this collaboration, Kogel says “analytics is very much a team sport, and every work stream is the result of a collaboration between a data analyst and a business stakeholder.”

Defining efficiency

Once a team is in place and creating analyses, leadership will naturally want to quantify the value of work created, and the team's overall efficiency. But efficiency can be hard to pin down. We asked these analytics leaders how they define efficiency.

For VSCO, this comes down to the “hours saved in getting to an answer, or days saved in the development cycle.” An analytics team might examine the causes of a download rate reduction, or the best drivers of retention. The answers to questions like these often reveal opportunities for the Product team to create a better experience. Kogel says that “without analytics, PMs or Growth managers would rely on inaccurate data, or waste precious time exploring less optimal routes.”

Efficiency can also be defined as a reduction of the time it takes teammates to get from question to answer. To this point, Kogel mentioned that VSCO's “self-serve systems and automated reports make data accessible to business stakeholders, so we can save time in the collection and production of insights.” While the team Patreon doesn't necessarily have a rigid definition of efficiency, Church agrees that the value of self-service systems cannot be understated. “If a teammate asks us a question, what percent of those questions already have an answer in a dashboard, and how easy is it to find that answer?” Analytics teams that invest in process automation and self-serve systems set themselves up for success by eliminating repetitive work, and free up analysts' time for more proactive work.

Measuring impact

Analytics teams need to surface data about their own work as well, as they need to measure their own progress and provide transparency to the entire organization. So we asked these team leads what metrics they use to measure efficiency and output.

At Fundbox, Gangal defined several measurements of his team's effect, which were relative to the function or department that originally asked the question. Of course, each organization has metrics that are unique to their vertical or goals, and each team uses analysis in relation to their own function. Marketing teams might look to a lift in funnel conversion, whereas sales teams are hyper-focused on improvements in churn and retention.

Gangal's team tailored their barometer of success directly to the audience that requested a given analysis. “Data science efficiency can only be measured by the impact it creates on business metrics”, Gangal says. “While it changes from case to case, that is the only true north.”

While Fundbox defines success by the metrics of stakeholders, at Patreon, Church is starting to measure success by the rate at which people are answering their own questions. “We're beginning to think more quantitatively about self-service rate,” so that success might look like an increased use of centrally located, shared, and explorable analyses that already exist.

Building the team

The most reliable way to ensure an analytics team is successful is to fill it with talented teammates. So we asked our interviewees what advice they'd offer others building out their team. Church advocates defining not only the team's goals, but also how that translates into a day-to-day job description. “If you're still working in spreadsheets and need someone to build out your data infrastructure — state that! Companies who miscommunicate about the complexity of their data problems can end up hiring the wrong folks.”

Ruben believes the larger mission of analytics at VSCO should be as important in the candidate selection process as the specific skills that candidate brings to the table. “Clarify the role of analytics and how it should interact with other teams.” Every analyst should understand how they serve each team within their own organization, and should feel supported with ample resources. Throughout the interview process, Ruben makes it clear to candidates that they're part of a larger goal.

For Gangal, building a team requires leaders to think about the impact they'd like analysis to have on the company, customers, and their careers as a whole. This means that you shouldn't necessarily limit your search to candidates who strictly match your style or immediate job description. Instead, Gangal suggests that leaders seek out challengers: “Hire independent thinkers who will question the status quo.”

As an analytics company, we're thinking, writing, and building with data and teams in mind. The Mode Community is here to help teams to be successful — whether you're an analyst looking to join a team (here are resources we recommend), or you're a leader looking to source great candidates.

Related Articles

Get our weekly data newsletter

Work-related distractions for data enthusiasts.