Before answering this question it’s important to notice the Process Mining activity must also be viewed on the the business user standpoint and should not only be focused on the business requirements. That may sounds weird somehow but the way the business user (with his own resources, skills and expectations) is going to analyze or monitor his own processes can totally change the way the solution itself will be used (and potentially the solution itself and its architecture).

So we may have to consider at least two usages:

• The “Ad Hoc” Process Mining: this usage suppose some pre-built Dashboards and/or reports provided by a Process Mining Solution. That also means some design and development have been realized and prepared for use by the Process Mining Team. Most of the time these assets are not modifiable by the business analyst and can only be customized (by using filters for example). If a modification is needed, we must go back to a design-development-validation phase between IT and Business. 
• The “On Demand” Process Mining: This usage is more focused on Process Exploration. The idea here is to provide a solution and the process data for the business analyst. Nothing is preconfigured inside the Process Mining Solution as we can assume the user can do his own analysis himself. That also means the Process Mining has to be simple to use and flexible enough to enable this usage. About the data, they have to be checked and prepared beforehand so that the business analyst would just have to pick it up without thinking about Data quality, etc. 

This two differents usages highlight two importants points:

1. The Process Mining solution can be different (rare are the ones which cover both needs correctly) as being able to create complex reports and dashboards can be incompatible with providing some easy exploration facilities.
2. The need to have data prepared:
   • For the “Ad Hoc” needs: Temporarily we can have a Process Mining solution which manages the data (integration & Quality), but that would be a good practice to create a separate storage for the data if the business plans to industrialize the dashboards and reports (mostly the case).
   • For the “On Demand” needs, it’s mandatory to create a separate storage area (or staging) in which we could check and prepare the data first (the Process Warehouse). The business users will then just have to pick the data needed from this area. Solutions like Blue Prism Process Intelligence for example proposes such area like this (the repository).

The schema below shows the On-demand analysis principle. We can see on the right the business user (aka the Citizen Process Miner) who can pick up the data needed for analysing his own process. In a previous stage these data have been qualified and prepared beforehand by the Data Experts. Of course this Data pre-work is totally reusable as the Data Expert have build reusable sscalable Data Pipelines.

One of the biggest advantage of this approach is the agility:

• By building Data Pipelines all the Data qualification and preparation is totally reusable and flexible
• The business Analyst can also (see the schema above) enrich the prepared data (which come from the entreprises systems and/or application) by adding his own informations. For example why not adding some reference with a new list of events ? See §The Step Name dictionary workshop
• The separation of duty between the Data Experts and the Process Analyst is also a key element as this enables each of these two different skills to focus on their expertise. The Staging area which is in fact the Process Warehouse seals the interfaces agreement. This can be a key success factor as many of the Process Mining projects failed because the solutions used did not proposed such separation like this.