← Back
Blog

Realities of Agile Analytics

September 3, 2022

Analytics teams can be partially agile while accepting the reality of deadlines and unknown data.

Managing analytics projects is hard. Teams face a mounting backlog while serving business stakeholders who are looking for a solution and  timeline ASAP, regardless of how messy the data they’re providing. Many analytics teams look to agile concepts to help adapt to changing business needs and uncertain data. 

  • Agile for Analytics - Can an analytics team be truly agile? 
  • Managing Deadlines - How to set expectations with the business. 
  • Portfolio Management & Scaling Agile - How to manage agile across analytics functions. 
  • Hub & Spoke Self-Service - How to enable customers and coordinate efforts across teams.

In the field, we see analytics teams face various challenges as they apply the theory of agile with the reality of analytics work. In this guide, we’ll walk through these challenges and how AlignAI helped Worthington Industries, a global steel manufacturer, overcome them. We worked with their Business Intelligence team to optimize their project management process, increase their throughput, and hit business deadlines. We’ll also highlight how your team can apply similar solutions today. We offer courses to help your team and would be happy to answer any questions, just email us at info@getalignai.com

We’ll be using a lot of agile concepts in this guide. If you’re  not familiar with agile, watching this video series may be helpful. 

Agile for Analytics

Working with data is a messy business. Many analytics teams start using agile processes to align with software and gain the flexibility agile promises. They take the classes, do the standups, and try to run sprints. But teams often lose momentum and devolve back into a free-for-all or revert to managing each project as its own effort. 

For your team, try the following steps:

  1. Create cross-functional agile teams that have the right skills to complete projects independently. 
  2. Set up a Team Backlog to plan and track the work.
  3. Define the types of work between use case development, bugs, support, and ad-hoc requests.
  4. Begin planning and executing sprints using the agile meetings of backlog grooming, sprint planning, sprint demo/reviews, and retrospectives.

Data science and machine learning work, especially initial exploration and model development, may not lend itself to the agile sprinting process. More traditional waterfall and phased approaches may be better suited for the research and development nature of the work.

Managing Deadlines

In order to plan their strategic initiatives, stakeholders need to know when they can expect their dashboard or solution. Those predictions are hard enough without all the interruption work, data quality issues, and other confounding variables at play. Agile prioritizes planning for flexibility over deadlines. This can create uncertainty for analytics teams as they service the data needs of the business. There’s also the inevitable dependencies and interruptions like data, system, and stakeholder access to navigate around. Agile requires velocity to make accurate predictions of deadlines. Velocity measures the work expected to be completed with each sprint and what is actually completed in each sprint.

Being data-driven, Worthington focused on the data they needed to make accurate predictions of when projects would be completed and measure the accuracy of those predictions. AlignAI helped their team manage expectations and timelines by implementing tooling and processes to track and forecast work. The team moved to Jira to organize and manage their work, making it possible to collect and analyze key metrics like velocity and time to business. Armed with clarity around velocity, the Worthington team can now more accurately predict deadlines and utilize retrospectives to adapt when they are missed. 

For your team, tune in your capacity planning by doing the following: 

  1. Define what types of work are in and out of scope for the sprint estimation process and team velocity (e.g. are support incidents estimated?)
  2. Set time expectations for how much will be spent on sprint work vs. unplanned/support work. (e.g., 60% of time on sprint work, 20% incidents, 20% meetings/ vacation.)
  3. Define how your team will estimate the work by picking between story points or hours. Hour estimates are a great way to start, especially if the teams are not well defined.
  4. Track team velocity for each sprint and focus the retrospective meeting on why the team did or did not complete the planned velocity for the sprint.  

Self-Service and the Hub & Spoke Model

Most analytics teams are looking to enable their customers to self-serve analytics to avoid building out dashboards. This is a massively impactful strategy that teams use to transform the broader organization to be data-driven. However, it creates new headaches from a project and work management perspective. Now the project management function needs to expand to coordinate work across hub and spoke teams, including all the dependencies, and avoid duplicative work. The centralized hub needs to understand the analytics initiatives of the spokes to curate the right data in the warehouse in time for those initiatives. The complexity of scheduling and prioritization grows. Project management plays a key role in realizing the change management component of self-service analytics. Organization, prioritization, and planning help the business teams utilize self-service capabilities and helps the analytics hub quantify the impact. If you can measure it, you can manage it.  

Worthington is driving initiatives to open up access for analytics and data across the organization. AlignAI helped their team structure the tracking of initiatives and plan for the dependencies, like making data available in the warehouse. We also focused on how to quantify the impact of the data enablement efforts as more analytics happens across the organization.  

For your team, do the following to enable self-service:

  1. Define the metrics to quantify the success of data enablement (e.g., number of dashboards published outside of the hub team)
  2. Determine the current baseline of the metric defined in step 1. 
  3. Collect a list of initiatives inside the spoke analytics team.
  4. Curate documentation and guides on the standards and learnings from the hub around analytics project management, business intelligence, and data engineering. 
  5. Train the spokes on standard processes and best practices. 

Scaling Agile for Analytics

Analytics work naturally requires a blend of skill sets and teams to bring together the solution. Data engineers, IT system owners, business analysts, data analysts, database administrators, and data scientists may need to coordinate their efforts to solve a business problem. They may not be in the same agile team or even managing their work with agile. This requires teams to implement portfolio management and cross-team processes to predictably deliver results. The scaled agile framework helps software teams address this problem. This approach can be overkill for most analytics teams, but there are useful concepts to leverage. 

Worthington focused on defining the life cycle and reporting needs at the portfolio level. AlginAI helped implement those data capture and reporting systems in Jira to understand return on investment (ROI), balance of work across requesting business units, and dependencies across teams.

For your team, do the following to help scale project management at the portfolio level:

  1. Collect and define all the initiatives or projects your team is working on in a portfolio backlog.
  2. Define how to capture impact or ROI and level of effort (LOE). 
  3. Determine how far out the teams will plan their upcoming sprints (typically 1-3 months) to synchronize dependencies.
  4. Build dashboard and reporting to communicate key metrics, including LOE planned for each business unit or project type.

Upgrade Analytics Project Management Today

AlignAI focuses on tactical process improvements around four critical areas – one of which is analytics and AI project management. We understand the importance of education and training for effective application, which is why AlignAI Academy updates content regularly and customizes learning paths. Whether you’re looking to leverage agile work management in analytics projects or properly manage analytics and AI implementations, we have a course for you. Check out our course catalog or contact us to learn more about the AlignAI Academy.