As our goal now is to understand the highest ROI actionable opportunities. With a more manageable process map, we can now flip through the various displays for insights..
In the duration view, we see the duration for each activity and we can clearly see that the invoice approval activity is the most time consuming, taking over 30 minutes on average..
In the rework view, we see that both request and approval have a lot of rework, which means they're inefficient and right for automation. Finally, from the finance view, we can see the resource hourly rate.
Which came from our HR system ingested during use Case 1 where we're able to see that activities relating to approval incur the highest costs. To better understand automation opportunity,.
Represent the automation rate. The automation rate metric considers all system resource activities as automated and activities of individual users as not automated..
Once the custom metric is created, we can bring up the custom metric displayed to visualize the automation rate within our process map. We can see that settlement and invoice creation is partially automated at 47 percent..
Approval activities are 0 percent automated, and payment processing is 100 percent automated. Next, through the inclusion of the resource cost data, we can calculate which activities will have the biggest impact on reducing cost if automated..
We can create a separate metric to measure the potential, or ROI, of the automation. This metric considers the duration of manual activities performed by a user and the hourly rate of that user. From this metric, we'll be able to see.
The potential savings for automating each activity in the process map view. We see that the approval activity in the process indeed has the greatest payoff for automation.
Posts Related:
- Python Part 13 - Error-trapping Office
- VBA to filter using clipboard data. Simply copy data and execute the utility. Time saving
- Design A Single Page Static Website (Custom Blogger Homepage)
At 438k for the six months of process data,
Or 73k per month. While the value provided is a great starting point for us to understand ROI, it is merely an estimation of that individual activity.Without considering how it may impact the process as a whole. To ascertain a more realistic ROI that incorporates interactions across activities in the process,.
We can use the process simulation feature. In the simulation feature, we can create a new simulation to assess the impact of automating the approval activity. Let's look at how it works..
We will use the original distribution of the data against the entire available date range. We set the cost attribute to the resource hourly attribute we used previously. In Resources, we can tweak the cost for each resource..
We ensure that system cost is minimized at one. In the BPMN model, we modify the simulation parameters for work distribution of the approval task so.
That system does 90 percent of the approval, which is what we hope we will get to once automation is in place. In the results of the simulation, we can see that the original process cost.
Was about $53 per invoice. By implementing automation for approval, the simulation feature shows that the resulting cost per invoice is now only $9. Not only that, we also have.
An almost 50 percent reduction in case duration. This results in a net of 71k in savings per month which is slightly lower, but aligns with the 73k we got from automation potential metric..
This savings is even greater than the ROI estimation from the COE tool kit but how do we get started with automation? Copilot recommends creating a Cloud flow for approval. By clicking on the suggestion.
We're taken to the natural language to flow
Creation experience where we can see what Copilot generated for us as suggested flows. This is a good starting point, but Copilot does not have the detailed information on how.The approval is conducted and neither does the process analyst. For this, we need to involve the business users and task mining to understand the approval activity on.
A more granular level to help us with automation creation. Let's start by using power automate process, minings task mining capability, where we'll be able to learn.
The manual interactions and applications involved for settlement approval. After creation, I can either create a new recording myself or open the process details to share the process with others..
We can share the process with other business users as contributors who can only add recordings and nothing else or co-owners who can assist in labeling recordings as well as view analytics. Business users will create.
New recordings using Power Automate desktop, available out of box in Windows 11. The easy to use recorder will capture the details of the desktop interactions, including interaction type, the UI element,.
The Window, the app, and other metadata for each action. As we add more recordings to the process, we can create the business taxonomy for approval so that there will.
Be alignment across recordings. Once we complete the required recordings, we can open each recording to see that actions are auto grouped using AI grouping and labeling functionality..
We can click on each action to see a detailed description as well as a screenshot to get a good understanding of what they did, including what seems to be a bespoke legacy invoice app, which will likely require RPA to automate..
We can also edit the grouping, like changing the AI recommended group name to the ones that we created in the business taxonomy. Once we are done with recordings, we can analyze the process to generate the process map..
In analytics, we can see the sequence and details of this process and how they diverge across the team. First, the approval request is open either from Microsoft Forms or e-mail. Then the settlement document is open..
Then the invoice data is extracted from the legacy invoice app. Settlement data, and invoice data is matched with the entries in SAP. Finally, after approval notification.
Either happens in e-mail or Teams. From the process map, we see that there are some variations in the order of activities, as well as some variants using e-mail instead of Forms and Teams,.
Which can certainly be streamlined for automation. Not only are we able to see the steps, but we can also switch to the application view, where we can see the usage of various applications captured,.
Including application switching, where constant switching between multiple applications signal some repetitive action like copy and paste. Our task mining solution even.
Includes an AI powered recommendation engine, to use all of that recorded metadata and recommend power automate connectors that can be used in automation. We see recommendations for each activity,.
Including Forms for getting the invoice approval requests, SharePoint for extracting the settlement document, SAP ERP for the matching step, Teams for notification and.
Desktop flows for steps that require RPA, like the legacy invoice app. By incorporating the invoice approval flow that Copilot recommended with the specific actions from.
Task mining and AI connector recommendations, we're now able to create the full automation flow for settlement approval. We see the flow triggers whenever a new approval is submitted in Forms..
We use the pre-built document extraction AI builder model to extract data from the settlement PDF. We get the relevant SAP invoice using the pre-built SAP flow templates as a child flow. Finally, we match the data across all sources..
If it matches, the flow will update the invoice as approved in SAP and post the message in Teams to notify payment team of auto approval. The pre-built flow templates were added after.
Power Automates acquisition of clear software and really simplifies a lot of the complexity for interfacing with SAP so you don't have to build all of that complex logic yourself..
With the flow created, we can test the automation by creating a new form request, which will act as a trigger. We can enter a sample invoice request and go back to the flow to see the execution of actions in real-time..
We were even able to use the machine feature to run the RPA flow interfacing with the legacy invoicing app on our desktop machine, where the app opens and automatically selects the corresponding invoice to.
Get the data before it returns. Finally, after matching and automatically approving an SAP, a notification card will automatically show up in Teams if the amounts match..
From testing, we see that this is much more efficient and runs in just a few minutes compared to the manual approval process it is replacing, which takes over 30 minutes on average..
Only Microsoft is able to provide this level of tight integration from process and task mining to AI recommendation to low code solution creation of apps, websites, and automations with Cloud flows or RPA..
