Close this search box.
3 Ways to Successfully Develop Agile Analytics

3 Ways to Successfully Develop Agile Analytics

They want a wilderness with a map—

but how about errors that give a new start?—

or leaves that are edging into the light?—

or the many places a road can’t find?

(“A Course in Creative Writing” by William Stafford, lines 1–4)

I was writing user stories for a project focused on delivering analytics dashboards, when these lines by the poet William Stafford came to mind. Most successful IT projects bring order and logic to a chaotic world. They make businesses easier to work for and with. They make business faster (“Speed is the new currency of business,” as Marc Benioff says). 

Analytics, or data science, gives organizations something a little different; it brings to light the invisible parts of a business. Poetically speaking, analytics brings insight from “the many places a road can’t find.”

At this point in enterprise IT history, most organizations have at least a basic understanding of agile development methodologies. What was once a novel approach to developing IT has become the standard. A generation of new project development team members has never touched a waterfall project. 

User stories, sprints, scrums, and so on have become the default methodology for delivering IT goodness to end-users. The reason for this is agile focuses on delivering high-quality, working software to users.

Improving sales, service, and marketing functionality

In the Salesforce world, we have seen the platform evolve from a great CRM system to a CRM and an application platform, to being a CRM, analytics application platform, and an analytics engine. 

Analytics! Once the domain of Excel ninjas and data scientists is now front-and-center on a platform that has brought sanity to sales, service, and marketing functionality.

Customizing and configuring Salesforce CRM fits well into the agile methodology. A business analyst discovers the needs of a user persona. Analysts translate those needs into user stories. Teams often use tasks to complete the user stories. Acceptance criteria ensure the solution matches what the story demands. 

User stories are put into sprints. The team completes the work. They report their progress (or blockers) at stand-up meetings (which, in the era of the Zoom meeting may be an anachronistic term). When sprints are complete, users get the value of the new functionality. Since the advent of the Agile Manifesto, versions of agile methodology have proven its value to the IT world.

There is one problem I am starting to see. That is, the agile methodology is a bit of an awkward fit when it comes to delivering data science.

Analytics: Navigating a wilderness without a map

Why is analytics different? The hope underlying analytics is that data discoveries will affect the continuous delivery of business and its strategy. It’s not about working software (we assume the software works); the focus is on discovery and insights. 

The act of discovery means that the process is never done in the same sense that a piece of software is done. For stakeholders that have a traditional understanding of IT delivery, and therefore expect an agile approach, this can be frustrating. Here are three ways to head off that frustration and reap the full benefits of agility.

1. Set expectations upfront

When it comes to delivering data science and analytics, over-communicating is your friend. For a lot of stakeholders, the whole concept may be new. Standup meetings for traditional software development can give a rough percent on how much is done. A developer might say something like, “I have finished the code, I need to write one more test script and this will take four hours.”

For the data science people, the end may not be so clear. Communicating how the project will be delivered and what they should expect to see will go a long way in alleviating frustration and allowing all stakeholders to give thoughtful input and ask good questions.

2. Establish clarity on data sources

In larger organizations, there are often myriad sources of data feeding the efforts of an analytics project. Engaging the owners of this data, ensuring data quality, and giving the stakeholders regular updates on any blockers in getting data needs to be a priority.

3. Celebrate progress and give updates on observations

In analytics projects, there’s not always a sudden moment where everything comes together. But still, there are items to celebrate. 

Stakeholders benefit when they hear observations about a dataset. What have we learned so far? Is the data conforming with conventional wisdom, or is it challenging commonly held beliefs? The stakeholders need to be updated as the project goes on.  

But what about when the conclusions aren’t exactly exciting? How about “the errors that give a new start”? Dead ends need to be celebrated for what they are: discoveries that can move you in the right direction. That’s why agile project management in the development process is key to a project’s success.

Learn more about our services and how we can help amplify your analytics capabilities.