Step 1: Recognising the need for collaboration and sharing
Many companies have a matrix organisation. Even more rely on a blend of in-house data and analytics resources, offshore operations and external business partners. In these scenarios, a number of key operating challenges emerge:
- Standards – are there common standards (such as data definitions, data quality, data governance) and shared ways of working?
- Collaborative development – are data projects and analytics briefs developed with close co-operation across functional teams, especially where offshore or external business partners are involved?
- Knowledge sharing – is there visibility across all practitioners of the work stack and does knowledge created around tools and techniques get shared between them?
- Connections – do data and analytics practitioners feel part of a virtual function, regardless of where they are physically located or whatever business division they support?
Data and analytics undoubtedly benefit from the “network effect” where the value of the service increases with the number of people using/delivering it. For some large organisations, the solution to the challenges above is to create a data or analytics centre of excellence (ACE or DACE) where all practitioners are co-located, or combined into a small number of grouped operations.
Yet this is not necessarily appropriate or achievable for all organisations for a number of reasons. A prime argument against centralisation is where analysts are embedded alongside their business stakeholders, either individually or as teams (and even functions in some cases). Organisations need to make data and analytics practitioners feel part of the business and having analysts embedded in the business is critical to how effective their outputs are and how well-aligned they are to business needs. At one mobile network operator, 60% of analysts were embedded, part of line of business meetings as well as having their own team meetings. Their presence alongside stakeholders helps because it means they adopt the same ways of working.
For offshore teams, communities can be a powerful way to help them feel embedded in the data and analytics function. They need focused and specific support at the same level as onshore practitioners, which one broadcast leader noted includes sending them presents.
In these circumstances, the right solution may be to establish communities of practice across the enterprise which unite data and analytics practitioners without centralising them.
Step 2: Defining your community of practice
Following a discovery process to identify and evaluate the range of data and analytics practitioners present in the organisation, the nature and scope of the CoP can be established. An immediate decision may be that multiple CoPs are appropriate to reflect the specific nature of the activities which each is tasked to carry out. In the example of one retailer, 13 different communities have been introduced reflecting the diversity of its business interests. Notably, there is an umbrella community which operates like a virtual centre of excellence. This is overseen by the group chief data officer whose role is described as that of “spiritual leader” to its decentralised teams. When additional pockets of data and analytics activity are discovered within the group, they are either assigned to an existing CoP or created as a new community.
Defining the scope of the CoP will depend on the resources and appetite within the organisation. One bank launched a virtual data science CoP via an email message to all staff, regardless of whether they were practitioners or not. It operates via a community site where ideas and examples are hosted, projects and challenges get proposed, and a community forum provides the “social glue”. One consequence has been to surface both the appetite within the business for the subject as well as data assets which may not have been identified during formal discovery procedures because they are operating within “shadow IT” environments.
Communities may be local to markets or specific business divisions or be cross-function and cross-enterprise. Simply establishing small CoPs which are then enabled with their own resources and a sense of identify can have a strongly positive effect on the practitioners within them. A more unifying approach can create a sense of common purpose, but does also require significant effort and resourcing – the retailer noted above runs a twice-yearly conference to which all 800 members of its CoPs are invited. Support for this was obtained via sponsorship from the company’s technology vendors.
Step 3: Components of effective communities of practice
To function effectively and to be sustainable, communities of practice need resourcing, support, engagement from leaders and stakeholders, and, ideally, a mission. There are some specific components that need to be in place in order to create a true sense of community. These include:
- Scheduled meetings – like any team, data and analytics communities work best when there are regular get-togethers, both formal and work-related as well informal and social. These can be weekly stand-ups and bi-weekly sprints, open to all members of the CoP, and may need to be timed so that offshore and remote workers are able to join. One telecoms network operator makes these “all-hands” meetings which everybody in the community is expected to join in order to share learnings and new outputs. Meet-ups and social gatherings should not be overlooked in creating this schedule. As one analytics leader put it, “never underestimate the power of cake.” Beer and pizza are known to have similar benefits…
- Internal communications – perhaps the most effort behind any CoP is keeping it fed with sufficient, fresh content so that members will regularly check-in for updates. It can take a considerable amount of time before a CoP springs into life organically with members actively sharing on their account. Until then, news about projects, success stories, examples of best practice, new hires and general “noise” about the activities of data and analytics practitioners need to be fed in. A forum is a useful tool to be able to offer as it encourages casual interactions and sharing. One organisation even places “Easter eggs” into its CoP communications – unexpected items of interest for those members who really dig into the content.
- Mandatory activities – within the world of IT, CoPs are a mature concept and many include mandatory activities as part of membership, such as core basic training modules and online refresher courses. Among participants at the roundtable, however, this was seen as a potentially divisive issue for data and analytics practitioners. One telecoms network operator had followed this approach and met with resistance because of a perception that any centralisation is a precursor to redundancies.
- Digital platform – while smaller CoPs may be able to interact and meet face-to-face, at any level of scale, it will be a digital platform that acts as the main meeting space. To be effective, this needs to be a separate solution from that used for project management, although there may be a crossover in functionality around collaboration, for example. Common choices for this are Bandcamp, Slack and Teams (part of Microsoft Office).
Step 4: Challenges and obstacles to the community of practice
Data and analytics practitioners are passionate about their domain and usually demonstrate a strong desire to deliver improvements in their business and often society as well. That ought to be a good basis on which to build a community of practice which harnesses this common purpose and enthusiasm. Nonetheless, CoPs do not necessarily thrive and achieve an organic life of their own. A number of challenges and obstacles can arise which may intervene. These include:
- Lack of collaboration – while individual teams may work well together, crossing functional boundaries, geographies or even ownership (such as where external business partners are involved) may not feel as natural or comfortable. This needs to be addressed through active leadership and formal methods of driving collaboration, such as meet-ups and mandatory activities.
- Lack of access – several members from regulated industries reported that their data and analytics teams in certain functions are not permitted to use cloud-based digital platforms because of regulatory constraints. While this can not be surmounted, ensuring there are sufficient face-to-face meetings or video conferences (which do not suffer from the same restriction) can keep the CoP bonds in place.
- Lack of content – there is common agreement that a key purpose of the CoP is to raise the visibility of what data and analytics is delivering for the business, to share best practices and common standards and to enable knowledge sharing. But these are not always easy to furnish since documentation of processes and data storytelling are not native skills. Providing training in data storytelling and presentation can build up this ability, while assigning content creation to a specific individual can ensure it actually gets delivered.
- Lack of focus – while it is tempting to view a CoP as mainly a social enabler, it must have a clear vision (such as aligning data and analytics to the corporate goal) and common purpose (such as developing common standards). Without this, it will lose momentum and also top-level support. Setting out this vision and purpose, then constantly communicating it to members, will maintain focus.
- Lack of outcomes – participating in a CoP should feel beneficial through learning and development. That means including knowledge sharing and skills development as part of what the community is focused on. Combing this with motivational activities, such as problems to solve or hackathons, combines all of the elements together effectively.
Conclusion
Recruiting data and analytics practitioners, then deploying them into the organisation brings a critical skills set to bear on the business. But developing this function and having leadership in place is not the same as creating a community of practice. A CoP represents a more formal, structured effort to build out vision and purpose which sustains involvement and improves knowledge and understanding.
Data and analytics can operate without being a community and often does, but it is also the case that higher staff churn can be experienced where individuals do not have a sense of belonging. For those with particular restless intellects and a desire for new challenges, such as data scientists, the CoP can become the fuel for their inquisitiveness that keeps them engaged.
While there is no handbook for running a CoP (other than this brief), a growing number of examples exist. Each reflects the nature of the organisation or the character of the leader who puts it in place, demonstrating that this is not a one-size-fits-all activity. The best way to find out what works is simply to get started.
