Sprinting through data projects

The key to successful projects can be failing fast.

Ginette LawSeptember 28, 2017
  •  data
  •  agile
  •  ux

In our last post, we talked about tearing down data silos through collaboration. In this post, we look at what Lean and Agile UX methodologies can teach us about sprinting through our data projects.

As we mentioned in the first article of this series, the three most common mistakes we see people making with their data projects are:

  1. Setting unrealistic goals because they’ve misunderstood the drivers of their target user / audience / market.
  2. Underestimating the extent to which local conditions and environments determine how, where, and why their target audiences behave in certain ways.
  3. Ignoring how these can change across regions / organizations / departments / industries / socio-demographics.

We’ve found that these can often be avoiding by applying Agile and Lean UX methodologies. One of the most useful methods is to integrate sprints into our data projects.

Avoid big and costly mistakes

While there are some differences between an Agile sprint and Lean iterations, the general idea remains the same: don’t wait till the end to find out if your project will fail or succeed.

Instead work in short iterative cycles or sprints. Fail early and catch mistakes before it’s too late.

Agile and Lean UX methodologies encourage you to validate or teste your ideas often and incrementally. You then iterate upon them with the feedback you’ve received. Sprints or work cycles can be as short as one week or as long as four. It all depends on the nature of the project, team, and the organization.

As Hilary Mason, founder of machine intelligence research firm Fast Forward Labs, points out, iteration is powerful because at the beginning of a data project it’s difficult to know if it will work or not. By nature, gleaning insights from data requires an exploratory approach. “Data science requires having that cultural space to experiment and work on things that might fail,” says Mason.

Approaching data in an iterative way makes sense. By being intentional, and always looking at where and why your data is failing you, you’ll learn to refine your data. You will can then make better decisions and take adequate action.

This approach also works when you’re developing a data-driven product. For example, while working with Knomos, a legal knowledge management start-up, we integrated Lean UX sprints into the company’s workflow.

Within two sprint cycles, we were able to help the company pivot their data-driven product. The user research and testing data we collected showed that their legal visual navigator didn’t meet their target audience’s needs. In less than four weeks, Knomos was able to refocus its product on a better audience, saving months of costly development work.

Fail fast and rebound

Sprinting or iterating on your data projects in short time-boxed increments will give your team the necessary time to explore your data while remaining focused. The goal should be to fail early rather than later. Identifying mistakes and deficiencies quickly will allow you to find solutions sooner.

So avoid the temptation to create a single, intimidating deadline. For better results and less waste along the way, envision a series of energetic 5k runs, not a marathon.

• • •
• • •