Best Practices Summary:

• Leverage a hybrid data analytics team with members distributed among functions to address specific business needs and a centralized core team to drive analytics standardization and scaling.

• Minimize distortions between systems without introducing excess rigidity by establishing where core alignment on principles is required and where divergences are acceptable.

• All teams leverage a single source of truth for underlying data to support end user solutions.

Case Study:

The customer loyalty program of a large technology company sought to grow customer engagement, retention, and revenue across consumer products. The program used data analytics approaches, including customer lifetime value analysis/segmentation and experimentation in product, marketing, and operations functions to learn about and then apply the tactics and customer personalization that best drove these growth goals. The team previously tried to find a single, perfect solution that provided the necessary KPIs and other insights to support both the operational rhythm of business and insight report outs, but despite working on it for multiple years the efforts all failed – no single solution ever hit the mark nor was adopted by business leadership. Instead, analytics experts were distributed across the organization’s functional teams while still working together to run experiments and share best practices.  The distributed approach helped address functional needs better, however the team quickly realized they were hitting a ceiling when they wanted to think bigger – for example, standing up new verticals or scaling experiments across the company.  Senior decision makers responsible for loyalty program-level investment decisions needed a common language for assessing data insights that matched how they normally made decisions, but the solutions the team had built were heavily focused on the needs of individual functions. 

The director of the loyalty program and his team recognized the need to stop constraining themselves to either a central or distributed solution and started again with a hybrid data analytics team and approach. In the hybrid approach, functional use case-level experiments were still managed within product spaces with distributed support, however when they identified the need to apply a solution broadly it was passed to a centralized core analytics team to scale and maintain.  The team didn’t sign up for everything, and this allowed for a narrower scope across channels and audiences while maintaining efficiency and accuracy. The team allowed themselves to decouple from perfection and identify two distinct sets of needs: a distributed context for supporting the operational rhythm of business, and an executive scorecard addressing customer lifetime value analysis with common business rules and language that were applicable across the company. The two solutions were built to meet everyone’s needs, paying close attention to how they connect in terms of underlying assumptions and business rules. A single source of truth was essential to ensure alignment across the distributed and centralized analyses, models, and outputs. Scrubbing the single source of truth data before it is used for either solution ensured it was consistent in the background, minimizing distortion of results. This ensured consistency where it was necessary and allowed for flexibility where it wasn’t. 

Through this more flexible hybrid approach, the business teams confidently informed their running of the business in product, marketing, and operations with insights. They were also able to meet with senior leaders to discuss and obtain additional investments using a common leadership language without the more detailed “baggage” that leadership did not need.  Trust in the data analytics team improved across all levels of the organization and profitability goals were more easily achieved through efficient experimentation and effective scaling of the best marketing and customer engagement tactics across the program.