Data Driven — How to Use Data to Supercharge Your Business
https://www.linkedin.com/in/patrickleebeatty
Patrick Beatty, Senior Director, Strategic Product Analytics, Naspers
According to PwC’s Global Data and Analytics Survey, 52% of executives that intend to enter a new industry, service, or product offering by 2020 don’t feel they can understand or manage their risks. Moreover, 61% of executives do not consider their organizations to be data-driven to help address these risks. These stark figures highlight the fact that in today’s business environment, if you aren’t data-driven, you are at a huge disadvantage relative to your competition. This plays out across organizations — if you don’t understand your customers, how can you build products to delight them and keep them coming back? If you don’t understand what interests your potential customers, how can you hope to market to them to get them in the door? If you don’t understand your unit economics, how can you effectively allocate resources? What is more, the impacts of being data-driven are even greater when you look at companies in emerging economies where systems may be less advanced and quality talent may be hard to retain due to a country’s “brain drain”. We talk with Patrick Beatty, our Senior Director of Strategic Product Analytics for his thoughts on creating a data-driven organization.
Q. We hear more and more about ‘data driven’ organizations, why is that?
A. It’s very straightforward — in today’s business environment, if you aren’t data-driven, you are at a huge disadvantage relative to your competition. If you don’t understand your customers, how can you build products to delight them and keep them coming back? If you don’t understand what interests your potential customers, how can you hope to market to them to get them in the door? If you don’t understand your unit economics, how can you effectively allocate resources? The list goes on. THE key question facing businesses of all sizes is how do you become data driven? Thankfully, the answer is a scalable solution comprised of three core steps: team, behavioral, and technical.
Q. Looking at the first of those, where do you start with building a data science team?
A. First of all, you have to get the structure right. Experience has shown me that there are too many ‘ivory tower’ data groups — and that just doesn’t work. It is essential to embed the data science team members across the organization, within the partner teams they are helping support as this keeps them close to the problems they are helping solve. However, while they are embedded, they need to report centrally to a technical leader, not within the partner team.
Once you have the right structure, you have to get your hiring right to fill the roles. The first hire should always be an experienced technical leader and mentor to lead the group, supplemented soon after with junior hires that can learn and have room to grow as the company grows. Across the board, the focus should be to hire for “scrappy”, collaborative, technical team members with a business and customer focus and desire to make an impact. What they do and can figure out is much more important than any school they have attended, so it’s vital to prioritize for a statistical and mathematical background requirement over business familiarity or a particular coding language.
Just hiring talent isn’t enough however, mentoring is critical to attracting and retaining top talent as well as running your organization well. Great mentors have the technical background to help grow their people, foster intelligent risk-taking, prioritize employee growth in a transparent and consistent way, adjust their style to their people, and actively set their people up to succeed and shine in impactful projects and work.
Q. So with your team in place, where do you focus next?
A. It is essential to get teams focused on the right, productive behaviors. Collaboration is the essence of an operationally strong company so that needs to be the first priority. One way to drive this home is the creation cross-functional groups (analysts, product, marketing, engineering, etc.) where analysts routinely collaborate with partners to understand priorities and impact.
Collaborating isn’t enough on its own. You have great communication, consistently provide reporting on how the company is performing towards its goals. Whatever you do has to be high quality, targeted and widespread, because while a lack of communication is often an issue, the problem can also be too much low-quality communication. So, while it is important to build new, quality tools to communicate insights, such as operational dashboards and predictive models that drive resource allocation, it is just as important to actively kill poor communications methods such as useless meetings and rote emails no-one reads. Good communication is clear, thoughtful, impactful, and actionable.
Underlying all this is a commitment to effective process from end to end. This starts with looping the data team into stakeholder initiatives right at the start to ensure success and time for to complete the project. Then you have to build effective team communications into the company’s ongoing processes — from the type and timing of mandatory team meetings, to standardized processes for follow-up on specific actions. The key is to set up processes so all involved know the purpose and outcomes for each part of the process so they can self-police the execution.
Q. What about the technical side of the equation?
A. The first thing I always focus on is data quality. “Garbage in, Garbage out” is not just a cute turn of phrase! Implementing quality data tracking and review has to be standard across the board, not just in more advanced areas like machine learning where clean data is essential. Fundamental sources of truth are crucial across an organization for it to run well. In addition, data needs to be standardized since there are many ways to measure things and each tracking tool may have its own definitions. The data and analytics teams should own the setting of common definitions to make insightful, useful, and actionable business decisions with all groups.
Next up is automation of data and processes to alleviate low-value work and increase efficiency and scale across the organization. Anything manual that is done repeatedly is a great candidate for automation, as is integration of data across systems. Even in Silicon Valley, it is surprising how much mindless work people do, when they could automate much of it away and use their skills on higher value-add work.
Once you have the right foundations in place, you can focus on innovation. Ultimately, encouraging and enabling intelligent risk taking and creativity to problem solve and using the appropriate tools is how data science teams deliver maximum business impact. The key is to empower people to feel safe to take risks if they have done their homework with evidence and can support their decisions. This allows speed in execution and autonomy that fosters leadership growth. Innovation is not only key to business success but to retaining talented team members.
Q. Any final thoughts?
A. If done well, becoming a data-driven business is almost certain to unlock your business potential and enable you to stay several steps ahead of your competition. Your business will be more agile relative to your competition, delight your customers, attract and retain high quality talent, and grow. Without being data-driven, you may still be executing, but without knowing whether your effort is going in the right places. Data science helps you crack the code!
