Workflow Intelligence - Key Activities in Workflows

Who can use this feature?
- Available with Workforce

Interested in a demo of Workforce? Contact us here.

Key Activities give a workflow context, transforming the raw activity Workforce captures into a digital twin of your process. By mapping the specific steps of a process onto a workflow, you can measure how much human effort it takes to complete, see where that effort goes, and prove the impact of process changes. This article builds on Analyzing Workflows.

This article covers the following:

Building Key Activities from Defined Events

Any process can be reduced to a start, a set of meaningful steps, and a finish. Workforce uses Defined Events for all three, which gives you full flexibility over which steps are mapped onto a workflow:

  • Start Event: the Defined Event that marks the beginning of the process.
  • Key Activities: the Defined Events that represent meaningful steps in between.
  • Completion Event: the Defined Event that marks the process as complete.

For example, to analyze a support ticket workflow, you might create the following Defined Events:

  • Agent assigns ticket to self — the Start Event.
  • Agent re-assigns ticket — a Key Activity you want to decrease.
  • Agent changes status to "on hold" — a Key Activity that impacts cycle time.
  • Agent applies Task Escalation Template macro — a Key Activity you want to increase.
  • Agent uses AI knowledge retrieval feature — a Key Activity you want to increase.
  • Agent clicks Submit as Solved — the Completion Event.

With Key Activities in place, you can surface opportunities to move your most important metrics:

  • Average Handle Time: reveal bottlenecks, time sinks, rework, and inefficient paths, and understand the impact of new tools and features.
  • Adherence: see where required steps are being skipped, or where undesired behaviors persist, and understand adoption of new tools and features.

Reading the Aggregate Flow

Any Defined Events you instrument automatically populate in the Aggregate Flow, as long as the events occur in the primary application associated with the WorkflowID (for example, Zendesk). The flow shows how work moves between events across all matching instances.

Note: Because all Defined Events on the primary application populate by default, irrelevant events can sometimes appear on the canvas. Toggle unwanted events off in the left panel to remove noise.

A flow like the one above can be interpreted as follows:

  • In the past 7 days, there were 116 tickets where a [Zendesk] Agent Clicks Submit as Solved event occurred (the view is filtered by that key activity in the left panel).
  • 22 of those tickets went directly from First Touch to [Zendesk] Agent Clicks Submit as Solved.
  • The median time for a ticket to reach [Zendesk] Agent Clicks Submit as Solved was 3d 3h.
  • In 5 instances, a ticket moved from [Zendesk] Agent Clicks Submit as Solved back to [Zendesk] Agent Clicks Submit as Pending.

Each node represents a Key Activity, each edge represents a transition between activities, and the counts and median times on the flow tell you how often and how quickly work moves between steps.

Measuring Mean Time to Complete

The most important question for most teams is how much human effort—hands-on-keyboard active time—it takes to complete a task or part of a process. This is the measurement at the heart of Workflow Intelligence, and it is the number teams work to drive down as they invest in process improvements and AI.

To measure it, choose a "To" event from the dropdown in the right-hand Workflow Insights panel. Workforce then reports the mean active time to reach that event.

Per-application attribution and path comparison

Once you have a Mean Time to Complete, two capabilities make it especially powerful.

  • Per-application attribution: Workforce breaks the active time down across the applications where it was spent, answering "Where does the effort go?" Many teams use this to consolidate work into fewer secondary apps.
  • Path comparison: filter to instances that took a particular path to see whether it changes the result. For example, add an Agentforce used event on the left panel to compare Mean Time to Complete for lead qualifications that used Agentforce versus those that did not, and prove the ROI of that investment.

This closes the loop described in the Overview: measure the process, make a change, then re-measure to see how the number moved.

Active time versus wall-clock time

Teams typically already know how long work takes in wall-clock time. Workflow Intelligence adds active time—the hands-on-keyboard effort—which is what process investments are designed to reduce. When a question really is about wall-clock time (for example, "How much time is added when an agent assigns a ticket to Compliance?"), include the relevant key activity in the left panel and change the median-time-from field on the canvas to that activity.

Note: A completion event can occur more than once in an instance. When you select a "To" event to measure active time, you can clamp the measurement to the first or last occurrence of that event. The default is first.

Key Activities in an individual instance

Clicking any node or edge brings up a list of matching instances, and clicking a row opens the individual instance view. With Key Activities in place, that view maps activities from all included applications onto the workflow—not just the primary app. This answers questions such as:

  • Did a given activity occur during this instance?
  • Where did it occur during this instance?
  • Who performed it during this instance?
  • How many times did it occur during this instance?

You can also set a "To" event in Workflow Insights on an individual instance to measure the active time a specific person spent before a particular step, or use the Key Activities control at the top-right of the canvas to answer targeted wall-clock questions.

Filtering workflows

You can filter your aggregate instances by Employee, Activity, and Time.

  • Employee: multiple selections are allowed. Use match on selected participants only to strictly filter to instances where all selected people were involved; otherwise the selection is treated as an "or" query.
  • Activity: behaves similarly to Employee. Key Activity counts also appear in the left panel, giving a quick sense of how often an activity occurred.
  • Time Range: filter to a specific date range.

Note: The Activity filter lists all Defined Events, but only events defined on the app associated with the selected WorkflowID will actually filter the instances shown. For example, when analyzing a Zendesk WorkflowID, a [Zendesk] event will filter correctly, but choosing an event from a non-Zendesk app will have no effect.

Note: The aggregate workflow map has a maximum of 1,000 instances. Above that, a random sample is selected to represent the aggregate workflow, and this is noted on the Flow button for transparency.

Best practices

  • Prefix event names with the app name, for example [Zendesk] Clicked Submit. Without the prefix, it is difficult to tell events apart in the Key Activities list.

Frequently Asked Questions

Why does an event I added to the canvas show no data?

The Aggregate Flow only maps events that occur on pages carrying the WorkflowID as a page property—that is, events on the primary application. If you add an event from a different app, it may appear as an option but will not populate on the aggregate map.

What is the difference between Mean Time to Complete and wall-clock time?

Mean Time to Complete measures active, hands-on-keyboard effort, while wall-clock time measures total elapsed time. Active time is the metric process investments are designed to reduce. For the underlying definitions of active, focused, and visible-tab time, see Workforce Active Time.

How do I compare whether a new tool improved a process?

Measure Mean Time to Complete for the process, then add the relevant event to the left panel to filter to instances that used the tool. Comparing the two results shows whether the path made a measurable difference.


Was this article helpful?

Got Questions?

Get in touch with a Fullstory rep, ask the community or check out our developer documentation.