Tell me and I’ll forget. Show me and I may remember. Involve me and I learn.
– Benjamin Franklin
According to Tableau, data visualization is the graphical representation of information and data. It’s as simple as that.
But how relevant is it for business analytics?
Surprisingly, many tend to believe data analytics does not depend on visual interpretation, but we must disappoint you. Various success cases on process mining implementation show the need to re-think how we treat large amounts of corporate data. The competitive advantage gained from using visualization in data analytics is undeniable.
Using data visualization tools, one can quickly perform business analysis and identify trends and patterns. For this reason, comprehensive data representation (visualization) is absolutely crucial in process mining.
The main data visualization challenges in process mining
There are different approaches to visualization depending on the amount of data, but when we talk about visualization in process mining, we primarily think of big data.
We researched the challenges that most visualization software encounters, despite being a different industry, and came up with a list of the four main issues and uncertainties surrounding visualization. Although, since we’re talking about process mining, let’s see how these issues affect our tool and what the ProcessGold development team does to solve them.
Challenge 1. Data – do you want it to be pretty or do you want it to be real?
The majority of time in visualizing a process is spent on data processing. Before visualization even takes place, an organization should make sure they gather the right information according to the objectives of the process mining project.
If the data is not gathered correctly, information bias might occur. Besides, when trying to create an outstanding analytical tool design, people tend to pay more attention to how it looks, rather than the results a software gives – therefore, adjusting the data to achieve a better-looking process.
Our solution – focus on raw data
Obviously, priority should be given to raw data – this is the data you can really trust to base business decisions on. Speaking from experience – for our team, it was challenging to build a new application for every request. The solution engineering team spent countless days and hours figuring out the perfect visual for each piece of the dataset.
Today, as a result of years of experience and developed templates, the actual visualization, which includes building up the dashboards, takes us less than 10% of the entire project time. We now have more time to focus 90% of our efforts on data, in order to keep it as clean as possible.
Challenge 2. The dangerous wish to stand out
Related to the previous point – imperfect data tools allow you to create visuals quickly, but lack precision and detail. Depending on the types of insights, an analyst requires more or less concrete data. Over complicating the visualization may lead to inefficiencies in data interpretation or even loss of information. We are pretty sure you don’t want to deep dive into bottlenecks displayed like this.
Such graphs are difficult to understand and require extensive analysis. Indeed, one easily gets lost following one line after another, trying to understand correlations, colors, shapes and words. Using this approach in process mining would result in quite a chaotic dashboard.
Therefore, it’s now more crucial than ever to focus on simple and comprehensive visuals. Try not to get caught up in everything the market has to offer, as they are often lacking in usability.
Our solution – Classic graphs and interactivity
At ProcessGold, we utilize comprehensive, standard graphs that display only the most important and relevant information. Depending on the information displayed, we use one or a combination of the following graphs:
These are displayed both horizontally and vertically, and are used to showcase the current situation, identifying the highest and lowest values. Overall comparisons also fall under this type of visualization.
This is the best way to track trends and monitor performance over a period of time.
This is another very simple and traditional way of displaying the most important information, which is visible as soon as a user opens the platform. Organizing data in a structured way gives you a quick update on KPI’s.
Lists are a simple yet efficient way to display a detailed story of events with a possibility to deep dive into each of them.
The masterpiece of ProcessGold is the classic process graph as well as its social network variation.
To address concerns around the traditional DOT algorithm, we created the TRACY layout algorithm that maintains a stable process graph layout and allows users to easily track changes. The whole idea of the algorithm is to focus on key activities, keeping them stable over time and throughout changes, assuring the vertical position of the process flow – the way business users tend to draw processes on paper.
The social network variation provides full transparency on teams and roles to inform workload schedules and performance. This allows you to identify best in class performance across teams to establish benchmarks and optimal locations for activities.
The Gantt chart works best for the presentation of single instances. For example, in the Gantt chart below, you can see that in this instance, the final check of an invoice was checked more than once.
Challenge 3: Showing everything and nothing
To be honest, it is quite complicated to make visualizations for a tool like ours because of how detailed and advanced it is. Due to its flexibility, the customization possibilities are almost limitless which is why it takes a while for new users to learn it.
After all, the displayed data should serve its purpose of providing the information clearly and concisely, and for this it’s important to highlight the important parts and make it understandable. Breaking the extensive data down into smaller topics makes it easier to understand.
Here at ProcessGold we try to understand what customers want to see and limit the information displayed, at the same time giving users a chance to interact with the visuals and discover more information. In fact, we were inspired by the details on demand feature mentioned in “Now you see it” by S. Few. According to the author, the feature allows the details to become visible when we need them and to disappear when we don’t, so that they don’t clutter the screen and distract us when they are not needed.
We have implemented this method throughout the platform, and it is also what you can see in ProcessGold 18 with the process graph. There used to be a hint displaying a lot of details on a case or activity, which included around 9 metrics. We modified it and created a simple hint with 1 metric displayed and a possibility to click and move to a more detailed dashboard. It significantly increased the ProcessGold platform performance, speeding it up while making the process graph more comprehensive.
Challenge 4: Change implementations resulting in a different visual
This issue is mostly related to the process graph – often after changes are implemented, the shape of the process changes significantly, making it harder to track the outcomes of the change.
In a perfect visualization tool for process mining, ETL (extract, transform and load) and data preparation should go hand in hand with visualization, which changes responsively according to recent data updates. If you actually do your data preparation in the same platform as your analysis, you will see changes immediately. Which is great. And that’s what ProcessGold has.
Having nicely displayed data in a graph is great, but it loses it’s meaning as soon as the dataset is updated. Thus, aligning the updates throughout the data and visuals is crucial.
What’s also important, is that you have an out-of-the-box template. You don’t want to reinvent the wheel over and over again. If you have a bar chart, then that should be the bar chart template you are going to use.
Since one of the key aspects of process mining is to deliver an overview of processes to business users who are not familiar with raw data analytics, visualization should be applied with usability in mind. When choosing a process mining tool, it’s important to ensure that visualizations provide precise, unbiased information in a comprehensive and structured way.
Your Content Goes Here
Your Content Goes Here