As data becomes more and more pervasive in workplaces, many executives are now realizing the value of having an analytics function to support their mission, whatever that is. However, some think that by hiring a few analysts or data scientists they have done all they need to do.
Building a great analytics function is not just about having the right people. It’s also about setting the environment up in a way that encourages high quality collaboration and great work. A big factor in that is the technology that is made available to the analytics function that enables them to carry out their work.
Here are some technologies that I personally believe are essential to have in place if you want to run a world-class analytics group.
Analytics work can be intense and detailed. It can also be complex and iterative. Analysts or data scientists need to be able to work together in a focused, uninterrupted way but also need to be able to communicate and share formulas, code, experience and functionality.
That’s all great if the team is co-located, but with team dispersion and remote working becoming more and more common, the right collaboration technology needs to enable this. Email and traditional messaging systems will slow teams down greatly because their do not allow the kind of sharing that’s needed here.
Tools like Slack or Circuit plug this gap by allowing collaboration in dedicated user-generated channels, meaning that analysts can easily navigate between different projects and have a linear record of prior communication on those projects. They allow easy communication of code in appropriate formats, and their screen sharing functionalities and voice communication capabilities are smoothly integrated into their messaging interface, allowing teams to simply jump onto a call or share an example whenever needed.
Agile management technology
World-class analytics team mustbe agile. It’s a necessity given the nature of the developing world around them. Projects can’t be effectively managed or delivered without appropriate progress tracking, and development needs to occur in short sprints to avoid wasted effort and inappropriate deployment of specialized skills.
Agile doesn’t just happen. It needs qualified people and enabling technology. teams should have at least one qualified Scrum Master, and all epics, stories and sprints need to be tracked and managed carefully and tightly.
Tools like JIRA, Trello or Planbox allow various forms of agile tracking, ranging from simple Kanban boards where requests can be logged, assigned, tracked and closed, through to the management of entire epics. Details can be captured as notes, including links to files or products, or code chunks. Analytics can be conducted which can provide information on delivery and turnaround, helping the team identify bottlenecks and efficiency issues.
Document and App sharing technology
Strong analytics teams should be developing tailored static or interactive documents to address key questions of the business, where possible allowing business clients to access data and analytics for themselves. Holding all work on an analysts local machine makes no sense, and that analyst’s talent will be wasted on repetitively answering the same question from the business again and again.
Data science publishing allows the rapid building of analytics documents or apps that can be made available to the broader business. These can range from simple views created for a very targeted client, to enterprise-wide reporting platforms. Documents are hosted on a server with controlled access so that users can log directly in and get what they need.
Tools like ShinyServer and RStudioConnect in R and Dash in Python allow data scientists to package their local code and publish it so that others can access it via the web. Built in reactivity allows easy construction of menus and filters so that users can help themselves to the data or analysis they need.
Any team working with large amounts of code needs appropriate version control. Without version control, a disaster is waiting to happen.
Version control tools like Github and GitLab allow teams to work collaboratively on code development without risk of version clashes. Work can be conducted in branches that will only be merged into the master version after a series of reviews and approvals. Strong integration capabilities with IDEs has made working in these platforms much easier and smoother recently.
Version control platforms also act as a central home for all the teams code. If the whole team is following good version control practices, this automatically means that none of their critical code is stuck on their local machines and it is all available for other members of the team to access — this is essential because code chunks from earlier development will be frequently repurposed in future development.
I believe each of these technologies are mission critical for any analytics team that aspires to high performance. As these technologies develop, they are also becoming more and more capable of integrating with each other. For example, a change that is pushed to version control can appear as an alert in your collaboration technology or your agile tracking tool. If you can build this kind of toolchain and environment then your analysts and data scientists will thrive and produce their best work.
Specific products are mentioned as illustrative examples only, and do not represent an endorsement of those products.