See what is the Most Common Path here
How can i find out the Most Common Path ?
Coming back to the stage 3 (Define) we asked the business analyst to draw down the Process Flow as it should be performed. As we had imported the data into the Process Mining solution we should be able to find out somewhere this famous Happy Path or something which looks roughly like this. Of course that almost never happens like this as the process reconstructed by the Process Mining solution is always far (and sometimes very far!) from the process expected.
Sometimes finding out the Most Common Path is just not something obvious!
So let’s consider these at least two options to detect what would be our best Most Common Path candidate:
Option 1: The first variation is obviously the right candidate !
By using a Process Mining Solution normally that would be pretty easy to find out the most common path (here below “Most Common Path” and “Happy Path” are the same as the Most common path is really linear and simple, without exceptions). This is the ideal case as we have a Process varation which covers more than ~30% of the Process usage.
Unfortunately, it doesn’t always happen that way all the time !
Option 2: We need to combine the firsts variations together
In some cases the Most Common Path is not so easy to find out as there are no Process deviation which covers +~30% of the process uses. However, by looking at the firsts deviations of the process it seems these firsts variations look really similar (with little variations). We can talk about little noise between the firsts devations which are not really relevant for a business standpoint.
So, why not combining these firsts variations to have an aggregated variation which could cover more than ~30% ?
By combining similar varations of the most used path could be a great idea to have our Most Common Path here. In the example above, the four firsts variations can be combined in one: the “Happy path”.
We can also note that most of the Process Mining solutions can combine the first variations to help in this work.
Note: This threshold (30%) is indicative and must not be used as a strict rule. It’s a good basis and this threshold can fit most of the time, however it always has to be challenged and why redefined by the business analysts.
We have found the “Most Common Path”
We find out a similar Most Common path for the process. What is now interesting is to see the level of use for this variation. By experience a good “Most Common Path” should have a usage between 30% and 85% of the Business Process use:
- Below the 30% threshold that does not mean we did not detected the “Most Common path” but we may have to consider some other criteria like:
- The steps repetitions. Most of the Process Mining solutions have the capability to merge the following and repetitive steps. Activating this option can help in this step merge.
- Some little variations inside the process itself which does not really change the process but create a new variation in the Process Mining Solution.
- Issue with the data : Maybe the 3 mandatory keys are not correct or there was a problem during the data integration.
- Above the 85% threshold, honestly that looks too perfect, but still possible especially if we have collected the data from only one application and there was no real Process bypass. That does not mean our analysis is correct as we may consider :
- Importing Process data from other data sources
- Increasing the scope of the Process data (data range, other dimensions, etc.)
There’s no “Most Common Path” candidate
We did not find a process or a variation which looks like the “Most Common Path” or have enough uses to consider it like this. That may also means there are too much variations in the process. In this case there are several options to consider:
- Maybe the Most Common Path drawn by the business user was definitely not the good one and we had figured out the real process execution. In this case we should have another variation of the process in this range [30% – 85%]:
- The Three mandatory key are not correct and must be reconsidered wisely with the business analyst
- The Data integration or pipeline must be reviewed to check if there was no issues during the import (generating for example a lack of data)