DEV Community

Cover image for Managing Power Automate Flows like a Genius
david wyatt
david wyatt Subscriber

Posted on

Managing Power Automate Flows like a Genius

Lets be honest, creating Power Automate flows is pretty dam easy (and I don't mean with Copilot), to the point where keeping track of them is.... well challenging.

Good news Microsoft heard you and created a very cool feature, that in my experience is still under used, its called Automation Center.

I suspect the Automation Center was created for Power Automate Desktop users, but that doesn't mean us Cloud foke can't use it too.

It has 7 areas:

  • Runs
  • Process map
  • Work queues
  • Errors
  • Desktop flow activity
  • Savings
  • Capacity utilization

Interestingly when I started writing there was only 6, then Errors snook in

I'm going to skip Desktop flows as its not my bag (if it is I would recommend following Agnius Bartninkas, so lets remove:

  • Desktop flow activity
  • Capacity utilization

So that leaves us with:

  1. Runs
  2. Process map
  3. Work queues
  4. Errors
  5. Savings

With these areas we can build a robust process to manage our flows across the delivery process.


1. Runs

With Power Automate run logs now available in Dataverse it means we can view that data a lot easier. The Runs section collates all of your flow runs in the environment, and shows it in 2 different views:

Overview

overview

This is your quick reference screen and where you should start your day. It quickly flags issues like failing flows, showing the detail and recommended fixes.

Below that is top like aggregated data about total and child flow runs, along with error rate, supported by a simple chart to show if its out of trend.

Finally you get list of the failed runs, here you can create a process map that shows the process flow between parent and child flows, and best of all it includes all input and output data.

failed run process map

Runs

A simple table of all of your runs with multiple filters that allow you to filter by:

  • Date Range
  • Run Status (Succeed,Fail,Cancelled,Terminat,Skipped,Queued)
  • Flows
  • PAD Machines
  • PAD Machine Groups
  • Trigger Type

run table

2. Process map

Probably the coolest part of the Automation Center, the Process map allows you to see a visual representation of a flow run. Not very useful for singe flow actions, but great for more complex parent-child flow and PAD.

It can be used to look at runs (like above in the Runs Overview), or just to view the solution.

process map failed

!process map](https://dev-to-uploads.s3.amazonaws.com/uploads/articles/aklz9s3m67iiisll91nw.png)

Unfortunately as its focused on a run it only displays parent - child flow (and maybe more with PAD), but I would love to see it show contents of a solution (fyi If you want to see visual representation of a flow check out Solution Explorer, demo here

solution explorer

3. Work queues

Work queues are mainly aimed at PAD, as they are designed to handle bottlenecks (in PAD its number of machines available to run, where cloud can run synchronously).

But there are still some good use cases for cloud, Ive gone into more detail about them in below blogs:

but the main two that are most useful in this circumstance is reporting and human in the loop.

Linking flow runs to actual pieces of value can be challenging, the default behaviour is to create a report, but another approach is to break up the value into items in your work queue, that way you can easily see the item detail and status.

The more useful though is human in the loop, the approval action is good for small amounts, but when you have lots of items the a work queue is perfect. All the items are in one screen, can be prioritised easily, and actioned.

The default screen shows a simple dashboard, that allows you to identify what queues have items, and the status (ideal for that human in the loop check).

main screen

You can also filter all of the charts by status, allowing you a quick view of potential workloads

queue filters

But what it really useful is you can click on the queue to get a quick link to all of the items for you to process

select queue

items

4. Errors

Errors pulls out all your failed runs into a easy to read dashboard, this is new and I still don't really use it, as I prefer to just use Runs and filter.

5. Savings

I've gone into a lot more detail about savings in a previous blog here, but in a nut shell in brings all of your savings for the environment in one dashboard.

You can flip between Time and Money results, and filter by time span and flows.


So why are all these areas useful for managing your flows, well it not only pulls all of the useful information in one place it also ensures you follow the right path.

My Process is:

Check runs

Im looking for failed runs to investigate, but also unusal volumes. High volumes could mean security issue, or impact API limits (each acount ahs 6k daily actions, 40k for premium).
On the flip side low volumes may impact ROI and make assumptions about future development work incorrect.

Process Map

The process map is the best first step for investigating failed runs. seeing a full view of connected flows allows us to see the inputs and outputs and identify the source (e.g child flo fails because parent passed incorrect data).

Work queues

This lists everything else that will have any sort of SLA (service level agreement) but in my world less critical then failed runs (generally business exceptions and human in the loop reviews).

Savings

Just like low run volumes if estimated savings are lower then expected I would look into resource planning for future projects using same estimations.

managing process


And that's how you master managing your flows, sadly there is no Automation Center for all environments, so you have to visit each environment that has flows you own/manage, but still its a fantastic tool and bubbles up issues quickly.

 
😎 Subscribe to David Wyatt

Top comments (1)

Collapse
 
duncan_true profile image
Dun

Super clear explanation—really appreciated how you laid out the daily routine for managing flows. The Process map section stood out, especially using it to trace parent–child runs with inputs/outputs for faster troubleshooting.