As much as I love cooking, I get even more excited about data visualization. Especially when it comes to visualization in process mining.
Big data gave the world an opportunity to dig into large sets of information and optimize daily life. A quick example – fast-food service is now even faster. Moreover, a large dataset that’s visualized correctly, becomes a story, rather than a set of complex numbers.
Unfortunately, I didn’t get to study visualization as a separate discipline during my Computer Science degree. However, I did study process mining, which in the academic world puts an emphasis on data, rather than visualization. The visualization in traditional process mining would be rather complex for someone without a technical background.
Generally, process mining implies working with large – and I mean really large – amounts of data. Therefore, you must consider visualization if you want it to be user-friendly and valuable. Luckily, I got a chance to learn more about visualization with ProcessGold, creating various applications for different clients with diverse requests.
Taking not only mine but the ProcessGold team’s experience into account, we figured that a good visualization is like a sauce you add to your dish – if you get the ingredients right, it will taste delicious. And again, it’s important to remember, that visualization (like any sauce) is supposed to highlight the taste of the dish (the data), rather than overpower it.
So here I am – ready to share with you my yummy recipe of the best approach to create nice visuals for your process mining tool.
To make your process mining visualization sharp and usable, you will need the following:
- Well prepared raw data
- Some adequate simplicity
- Communication skills and empathy
- Logical thinking
I believe you already have all these ingredients on hand. So, let’s start cooking!
1. Well prepared raw data – be true to yourself
Before even starting to prepare the data – think of what you want to analyze, what goals you want to achieve as a result of process mining. Visualization will depend on the specific information you want to see, since you simply can’t display everything at the same time.
People often tend to tweak the data, trying to make it more understandable. My advice: forget about it. If you don’t want to end up looking at unrealistic insights and making risky decisions, make sure you work with the original data and not the simplified version.
Remember that the outcome directly depends on the input. That’s why in ProcessGold, we try to keep the data as original as possible, leaving the real information for analysis. Although, it is a challenge – how can you display a lot of complex information in a simple and clear manner? – you obviously don’t want to overload the end-user.
That’s why we add another crucial ingredient – simplicity.
2. Keep it simple
I believe that for visualization, simplicity is key, and luckily, Stephen Few agrees with me. Quite often, people use outstanding, well-designed, diverse models and graphs to grab attention and satisfy user esthetics, but this doesn’t always give them the necessary result.
It is OK if a beautiful image is what you really want, but if you are going to use the information to make global business decisions – looks are not what you should care about most. Keep it as simple as possible. You may indeed find some visuals boring – it may be just another bar chart, but it is what will help you get the information quickly.
Indeed, data is not about being the most attractive if you want to work with it. The whole focus should be on the way people understand it best.
Seeking for simplicity, consider using interactive charts and graphs. This way you will be able to keep minimal information on the dashboard but give the user an opportunity to deep dive into any aspect, if necessary. In process mining, the downside of the interactive charts and graphs is impermanent structure after the change’s implementation. Therefore, the value of interactions is highly dependent on the layout stability. The user should be able to track the changes he implements and their influence on the process.
3. Communication and empathy – focus on the customer
So, you have mixed your real data with the simplified visuals. What’s next? The mixture lacks some knowledge on what customers prefer – what their tastes and particular needs are at this moment.
Along with simplicity, the trick of solving visualization issues is deciding what information to show. You need to have a clear vision of what is most relevant for a client to see immediately and what is better off hidden. It’s quite a common problem among any type of visualization tool – you want to show everything at the same time, but risk ending up with dozens of different overloaded dashboards, where no one can see anything anymore.
In the past, we often didn’t know if things would work and we’d just launch sets of functions, waiting for reactions. It then took us several months of internal discussions to figure out an improvement plan based on feedback. Indeed, it took us a while before we brought the improvements to life, desperately wanting to achieve the perfect state of an application before the next release.
Sounds enthusiastic, but we decided not to work like this anymore. Now, we pay major attention to usability testing, and I can tell you this is one of the most efficient ways to identify content for an application. Besides, we now adopt the terminology of the platform depending on the business, so that the end-users interpret the information in terms they are familiar with.
During the application development, if there is even a tiny doubt – we prefer to ask customers directly. Normally, they appreciate being heard and openly share their opinion. Most of the time it gives you valuable insight on improvements. After all – the platform is for them and they should feel confident using it.
Of course, sometimes there might be disagreements, but partnering-up does help identify the best way to dig out insights. Most of the time, your client knows what is required to do their job. Keep in mind, there is no point in creating extra elements they won’t use, simply because you find it necessary.
Using this method for one of our clients, we implemented a better SLA process visualization by breaking it into 3 sub-processes, resulting in a decrease in throughput time. As a result, two processes fell under the control of our clients and could be significantly improved, leaving the third process as the supplier’s responsibility. We tested 4-5 design iterations, before we identified the best one solving the client’s issue.
Now it can be applied to more SLA processes across various industries.
4. Think straight and use your gut feeling
If you look at the dashboard and get lost – other people will get lost too. Your mission is to not let this happen.
Even taking professional process mining academic courses won’t give you one generic advice on how to visualize the data, so use your head. Seriously, think of whether something you’ve created would work for yourself if you had to analyze that data.
Don’t overcook it. The final goal of any visualization is to quickly understand what is going on with the process. People performing business analysis and concluding strategic decisions normally don’t have time to stare at the screen the whole day. They appreciate efficient tools that deliver understandable information in a comprehensive way.
Add this to your mixture, spice it up with some creativity and care. Your visualization sauce is ready.