Luma Analytics Logo .png

Tel: +64 274 670912

Auckland Chancery, 41 Shortland Street, 


New Zealand, 1010

  • LinkedIn - Grey Circle

© 2018 by Luma Analytics Ltd. 

Is Your Analytics Team Structured to Win?

To compete in the age of the customer, companies are starting to truly grasp the latent value of their data and the need for a strong data science function. Whether it’s to predict customer needs, offer personalised experiences, drive operational efficiency or power digital transformation – data is always the foundation for success. As momentum and investment in data science grows, debate has now turned to building, structuring and fostering the best analytical teams to maximise results.

In this blog series, I’ll examine the key building blocks for delivering high performance analytics that creates a competitive edge. This first post explores the best team structure, further posts will cover talent and resourcing, culture, business partnerships and moving to agile analytics.

Choosing the right model

When it comes to team structure there is no simple, one size fits all model. Instead, it depends heavily on the size of your business, the level of analytics maturity, and your aspirations with data. The good news is there are some key principles and a few universal truths that in our experience drive the best team results.

Whether your building an analytics function from scratch, growing your current capability, or reorganising your existing teams the key decision you need to consider relates to the centralisation spectrum. The diagram below shows the range from fully distributed analytics teams to completely centralised models with the pros and cons of each.

Analytics Structure Options

At one end of the spectrum decentralised models offer flexibility and close alignment to the business, at the expense of efficiency and governance. While at the other end, centralised teams provide consistency of results and best practice knowledge sharing, at the expense of business area expertise and resource agility. Organisations will typically move along this spectrum as their business requirements and analytical maturity grows. They can also shift from one end of the spectrum to the other, usually triggered by a catalyst event like data quality issues leading to centralisation or poor delivery from a central team driving business units to decentralise and hire their own analysts.

Hybrid – the best of both worlds

There are certainly situations where fully centralised or decentralised teams work- but in our experience the most sustainable analytics structures are hybrid operating models. These are also known as hub and spoke, Centres of Excellence or squad analytics. They attempt to leverage the best of both worlds, by having analytics teams as close as possible to business stakeholders whilst at the same time reporting through to a centralised function to create strong data foundations, standardisation, best practice technical standards, and cross functional prioritisation.

Hybrid gives you flexibility to maintain distributed teams that directly support each business area whilst also developing core capabilities in your central hub. For example, BI reporting can be run and resourced from your centralised hub team. A single reporting team is better placed to produce consistent, accurate, automated and independent enterprise wide reporting. Similarly, Data Strategy and Data Governance programmes should be run from the central team to ensure enterprise coordination and strategic alignment. A core team providing robust data and reporting foundations will free your business insight and data science teams to focus on the proactive forwarding looking insight and advanced modelling the business needs to drive real performance gains. The diagram below shows an example of a hybrid model with core competencies in the hub while spoke teams focus on supporting business functions, projects and strategic initiatives.

Hybrid Operating Model

A big advantage of hub and spoke analytics structures is you get to build a strong community of practice for knowledge sharing, intellectual challenge and developing best practice technical standards. It’s important here to have strong leadership and governance from the centre to ensure analysts stick to agreed methods, otherwise pressure from the business can result in short cuts creating longer term technical debt. Analytics is, and always will be a technical discipline and the best data scientists and analysts thrive in an environment of continuous learning, challenge and problem solving. The critical mass created from hybrid analytical teams are a great way to create this.

In Summary

There are several design decisions when looking at building the optimal analytics function for your business. No one size fits all, and it does depend on the scale and analytics maturity of your organisation. However, over time you should consider moving towards a structure that leverages the best bits from both centralised and decentralised models. It’s a balancing act but hybrid models with the right talent and culture can transform your analytics function into a powerful source of competitive advantage and help you win big with data.

Stay tuned for more posts in this series to help create your blueprint for a high performing analytics function.