There is no one-size-fits-all formula to creating a data centric organisation, however D&A leaders identify a set of 10 common considerations that can help you define the right formula for your own organisation.
Data Leaders members can read the full post and connect with peers via the Data Leaders Hub.
Data and Analytics leaders know that creating a data centric organisation is not something that happens overnight. Far from it; it’s a process that evolves as their organisation’s maturity grows and business strategy changes. While there are structures and models that work better or worse in certain types of organisations, the key is really to understand your own organisation’s needs.
During a recent Data Leaders peer discussion, D&A leaders shared their approaches to designing the optimal organisational structure for where your company is now.
1. Understand your priorities and the capabilities you want to build in the next 2-3 years.
How you structure yourself will depend on your data strategy, the capabilities you want to build and where your focus is going to be of the first 2-3 years of your programme. Try not to think longer term: you can’t do it all.
2. Assess your maturity and identify gaps.
Conduct a data maturity assessment to highlight not only where you are but also where your current weaknesses are, for example in tools, people, culture; skills processes, policies and controls.
3. Expect the structure to evolve over time.
If you are just starting to build a data capability a centralised team now with a view to becoming a decentralised hub and spoke structure in three years’ time could make sense. You don’t have to commit to just one structure that will be set in stone.
4. Design your structure to fit within the existing structure of your organisation.
The best data organisation should reflect the company itself, for example, it should mirror decentralised or centralised organisational structures. Moreover, take into consider the company culture, (is it risk adverse, innovative?), strategy (customer-focused, high growth-focused?) and adapt to what the company can absorb.
5. Separate your physical structure from your delivery and operating model.
It’s likely that your physical organisational structure will differ from how you deliver and operate your data products.
For example, your physical organisation could be structured around data functions, (e.g. data management, data governance and data analytics) while your delivery model could be organised by business function or domain.
Once again, these models should not be thought of as fixed – especially the delivery model – as focus areas, use cases and strategy will change each year. Therefore it’s important to keep in mind that your organisation must be able to flex.
6. Keep the data with the business, but think about IT ownership.
While most D&A leaders dream of owning the end-to-end data and IT capabilities, in most organisations this isn’t possible. What to own and what to “give away” is a struggle for most leaders with no clear cut answer. One D&A leader’s best advice was “think about the politics and think about which functions you want to fight for and which you don’t. For example, maybe the data architecture and models live with me but I don’t need to lead the run of the platform.”
7. Expect to deploy a data literacy programme.
Investment in organisation and IT can all go to waste if people are unable to utilise the data products and solutions built by the data team, so expect to create a data literacy programme.
Ownership of the programme and how extensive it needs to be will differ wildly based on the company’s business activities, data strategy and data personas and could result in a company employing a head of data culture, or assign the task to data governance or data transformation roles.
8. When it comes to hiring data scientists, think about where you are in building your foundations.
Members have often said that their senior executives perceive data scientists as Swiss Army knives to be leveraged in every which way possible. All D&A leaders caution against this approach, instead advocating once again to align requirements to maturity, strategy and to assess the skills required to build the capabilities you want in the next 2-3 years.
Usually, this means focusing first on foundations and hiring data architects and engineers before data scientists; mis-hiring is not only expensive but disrupts productivity and holds back the company’s data transformation.
However, do plan ahead; look 12 months down the road to the skills you will need and start to put them in place.
9. When it comes to AI and ML, don’t run before you can walk.
While senior execs may be looking to incorporate AI to accelerate innovation and bring internal efficiencies, off-the-shelf AI is a more viable option than in-house AI until your data foundations are established.
10. At the same time, don’t miss an opportunity to enhance your offer and protect your business through proprietary AI models when the time is right.
If you’re providing your data to vendors for them to calibrate their own models, it makes better business sense to create the model, monetise it and sell it to others.
Ensure you get stories like this and many more interesting insights from data and analytics leaders like you – directly to your inbox, twice a month – by signing up to our newsletter. Would like to become a member? Get in touch!