09 Mar 2016

The 5 D Approach to Agile Analytics - How to Accelerate Technology Adoption

Tim Archer

*Blog image by Jakub Szepietowski

A New Challenge for Enterprise Analytics

A sharp growth in self-service analytics tools are changing the way people interact with data. When enterprise technology platforms cannot deliver, business users are now able to use self-service tools to build their own analytics applications from the ground up using agile methodology. However, simply knowing how the tools work does not guarantee successful adoption. In fact, rather than driving engagement, poorly built analytics dashboards actually pose operational risks to a business, resulting in applications that do not scale, proliferation of functional silos and failure to address end user needs.

That's the bad news. The good news? There is a simple way to ensure that dashboards connect with their consumers – allowing them to quickly decipher complex data and direct their attention to areas of concern and need.

The 5-D Approach to Agile Analytics

At Concentra, we adopt a five step iterative approach that allows us to connect dashboard development with end user needs, thereby improving the speed and success of embedding a new analytics process to a business.

Figure 1. Process Flow: The 5 D Iterative Approach to Agile Analytics

*Disclaimer: It is important to note that this process is aimed at delivering dashboards that are expected to be deployed to a wide audience as a solution for the medium to long term. It is not an approach that needs to be used for business and data discovery, which should not be limited to a framework!

1. Discuss

Take the time to talk to colleagues across the organization to fully understand their painpoints. The more people are consulted the more ideas and unique challenges will be uncovered. This discussion will drive stakeholders towards a consensus and create a feeling of ownership for the final solution. This makes it far more likely to stand the test of time and be fully embedded. 

Unearthing business challenges in the initial stages of an analytics work stream allows analysts to join the dots with other departments and processes that have an impact on analysis. Having a series of conflicting, disjointed dashboards creates confusion, rather than clarity. Communicating business insights through logical storytelling has been an emerging theme for years within the industry for this reason. 

2. Define

Having discussed and generated consensus, the next step is to understand what data can support the analysis as well as the business logic and calculations needed to answer the questions posed. The source data will have a large bearing on the analysis and will often determine the calculation and granularity of a metric. Let's consider a 'Sales versus Target' metric where:

  • Sales data is captured by item, location at the date and time at which the transaction took place
  • Target data is defined by the forecast which is created per location per month

In this example, the metric can only be calculated at the granularity of the highest aggregation of the two measures needed to calculate it. Therefore, the lowest level of detail a 'Sales versus Target' metric will be governed by the target data. Furthermore, calculation details  must be defined consistently, as there may be multiple ways of analysing a simple metric. In our 'Sales versus Target' example you may find that all of the following calculations are important:

  • Distance from Target: Target – Actual
  • Target achievement: Actual / Target
  • Target error : 1 – (Target / Actual)

Equally, at this stage it should be considered how often the analysis would need to be updated to meet the business requirements, ensuring that the source system is able to support this.

3. Design

By now, stakeholders should have a good grasp of the problem and data requirements to start the wireframing process. A wireframe can be done on a whiteboard or with a wireframing tool (such as Balsamiq) to facilitate rapid building of mock-ups. Whilst tools like Tableau and Alteryx allow for their own rapid discovery, at the design stage, we recommend to remove yourself from the tool as it can open up different ways of thinking and communication. Wireframing will help designers to work through the essential functionality versus the nice-to-haves.

Discussions around how a series of visualisations and dashboards interact with one another is key at this stage. Setting up a simple reporting schema will give an understanding of how different stakeholders may access different areas of the analysis at different points within a certain reporting time period. It will also allow a Birdseye view of the whole solution, making it easier to decide on which metrics to put together and analyze in a single view.

It is vital to stay engaged with stakeholders throughout this process. Start taking feedback and amend the wireframes accordingly. They should be used as the blueprint during the 'Develop' phase! 

4. Develop

This stage focuses on moving all ideas into the tool of choice. The logical place to start will always be the data, as this remains fundamental to any visualisations. Some calculations may need to be performed before the data can be used. Alteryx is a great tool to quickly iterate different approaches to analysing, blending and shaping data for use in a visualisation tool, even if it may not be used for the production version.

The data can then be used in the visualisation tool to start creating a working version of the wireframe that has been shared with the stakeholders. Typically, working towards a wireframe alongside the business user will give a better sense of direction and reduce the risk of being lost in the endless possibilities of analysis and visualisation options.This stage will see a lot of iterations between some basic visualisations and the data, which will ensure that the format of the data is suited to the visualisation tool.

5. Deliver

In the final stage, the preliminary visualisations that have been built in the Develop phase should be enhanced to increase their appeal to the target users. Applying a theme that either supports the organizations style or supports the story of the dashboard will greatly increase engagement. Further formatting will help the user to navigate and interact more with the analysis – simple calls to action like 'Click here to see....' or descriptive tooltips give instant feedback to the user and allows them to delve deeper for further insights.

This stage is by far the most important when it comes to improving the ability to make a dashboard with lasting impact throughout the organization. Upon delivery, the creator should not stand still. Constant evolution is important to keep the work current and adaptable to ever changing environments. Seeking constant feedback and iterating through the process will help to drive this. For more ideas to combine design principles with your analytics dashboards, read our blog on how to create engaging dashboards.

Next time you are about to start an analytics project, make sure to think through these five steps. They will help you to accelerate user adoption and ensure long-term usage of a commissioned solution.

Download free trials with our partner technologies:

Tim Archer

About the author

Tim Archer - Consulting Manager. I visualise data to help our clients see and understand their business better as a living! I enjoy visualising all sorts of data – particularly sporting and current affair data. My blog articles will usually pull together multiple sources of data to uncover interesting insights and offer different perspectives. No matter what the subject, they will always leverage visualisation best practices to present stunning insight to the masses.