Read workflow analytics

What this achieves

Analytics shows where approvals are slow and where they break, across every workflow you run. It is the difference between knowing that approvals feel slow and knowing which step is responsible.

Required role: Administrator.

Steps

Go to Approvals → Analytics.
Read the headline figures across the top.
Open Step bottleneck analysis to find the steps holding requests up.
Choose a workflow under Completion rate to see how one route performs.
Read Approver performance for how long individual approvers take.

The headline figures

Option	Description
Active workflows	How many routes are in service.
Running instances	Requests currently in flight.
Avg completion time	How long a request takes from start to finish.
SLA breach rate	The share of steps that passed their deadline.
Failure rate	The share of instances that failed rather than completed.
Overdue	Tasks past their deadline right now.
Escalated	Tasks that have escalated.

A high breach rate usually means the deadlines are unrealistic rather than that approvers are slow. Compare it against what the steps actually allow before treating it as a performance problem.

Example: if HC Corp’s promotion route breaches on almost every Finance step, the question is whether 24 hours was ever achievable for a step that needs a budget check.

Finding the step that is responsible

Step bottleneck analysis ranks steps by the average time requests spend on them. The slowest step is where to look first, and it is often not the one people complain about.

Rejection trend shows the share of decisions that are rejections over time. A rising rejection rate on one route usually means requests are being raised without something they need — which is a form or a briefing problem, not an approval problem.

Example: a rejection rate climbing on training requests at HC Corp turns out to be requests submitted with no reason written. The fix is upstream of the approval.

Approver performance

The approver table lists each approver with their average response time, how many decisions they have made, how many they approved and rejected, their approval ratio, and how many are still pending. Anyone carrying an unusually large pending count is flagged.

Read it as a load report first and a performance report second. A large pending count usually means too many routes point at one person.

Example: if every route at HC Corp escalates to Daniel Okonkwo, his pending count will be high no matter how quickly he works. The answer is to spread the escalation targets, not to chase him.

One workflow at a time

Each workflow has its own Analytics tab showing the same shape of figures for that route alone: total instances, average completion, breach rate, failure rate, instances by status, and its own bottleneck steps.

Below them, a node performance table breaks the route down step by step — executions, average duration, completed, failed, and failure percentage — which is where a single misconfigured step shows itself.

What happens next

Act on what you find in a new version of the workflow: adjust a deadline, add an escalation target, or remove a step that adds a wait without adding a decision. Requests already running stay on the version they started under, so improvements show in the figures gradually rather than at once.