1. What Insights shows, and who sees it
OpsBrain records everything that happens during guided execution: every step completion, skip, flag, note, and timing. Insights is where all of that becomes readable. There are two views:
- The organization-level Insights page (Insights in the sidebar): every signal across all workflows in one place, with date-range filtering and period-over-period comparison.
- The workflow-level Insights tab on each workflow's page: step-by-step analytics, force-skipped required steps, flagged issues, and worker notes for that one workflow.
Who can see what: the organization-level Insights page is managers only. It appears only in the manager navigation, and the route itself is manager-gated. On a workflow's page, the Insights tab is visible to anyone who can open the workflow, alongside Overview, Steps, and Runs; the Versions and Settings tabs are hidden from workers. There is no separate team-level or step-level insights page: team context lives on the team detail page, and step analytics live inside the workflow's Insights tab and the org-level bottleneck panels.
Before any runs exist, the page shows No data yet: once workers start running workflows, live analytics, bottlenecks, and improvement recommendations appear here.
2. Date ranges and comparisons
At the top of the page, pick the window you want to analyze: 7 days, 30 days, 90 days, YTD, or Custom with your own start and end dates, then press Apply.
Every KPI carries a delta badge comparing the selected window against the window immediately before it, labeled vs previous period. If there was no activity in the prior window, the badge reads No prior data instead of pretending a comparison exists.
3. The six KPIs
| KPI | What it measures |
|---|---|
| Total runs | How many runs started in the window. Your headline volume number. |
| Average completion time | Mean duration of completed runs. Watch this after publishing workflow changes. |
| Incomplete runs | Runs that did not reach completion. A rising number usually points at a blocking step or unclear instructions. |
| Completion rate | The share of runs that finish successfully. |
| Active workers | How many distinct people executed runs in the window. |
| Flagged steps | How many step-level flags workers raised. The rawest "something is wrong" signal you have. |
Each KPI shows its value for the selected window plus the vs previous period delta described above.
4. Every chart and panel, explained
Runs over time
Daily run volume across the window, broken down by terminal status. Hover any day for its exact breakdown. Use it to spot weekly rhythms, sudden drops, and the effect of publishing changes.
Status mix
A donut of how runs ended, with Total runs in the center. The status colors cover the full lifecycle, including sub-workflow states like handed off, waiting for sub-flow, and pending pickup. Click a slice to drill straight into Active Runs filtered to that status.
Top workflows by volume
The most-run workflows in the window (up to 8), each with completions, success rate, and average duration. Click any row to open the workflow. This is where you learn which SOPs actually carry your operation, and therefore which ones deserve optimization attention first.
Worker leaderboard
Who's executing the most workflows (up to 8), with run counts and average duration per person. Before activity exists it reads No worker activity yet.
Step bottlenecks
The steps taking the longest on average across all workflows (up to 8), each showing the step's position and title, the workflow it belongs to, how many runs hit it, and its skip rate. Bottleneck detection here is straight arithmetic computed from recorded step durations over the window, not an AI judgement, so you can trust it as a measurement. These rows are the natural candidates for redesign.
Most-flagged steps
Where workers signal trouble most often (up to 6), with the workflow name and the top flag reason per step. When the window is clean it says so: No flags raised in this window.
Flag reasons
A bar breakdown of why workers raise flags, across all six flag types: General, Unclear instructions, Equipment issue, Missing resource, Policy conflict, and System error. The shape of this chart tells you whether your problem is documentation (unclear instructions), environment (equipment, missing resources), or process conflicts.
Optimization activity
How the improvement loop is doing in this window: suggestions surfaced and acted on, bucketed as Pending, Accepted, Rejected, and Dismissed, with an Open inbox shortcut into the optimization inbox. A large Pending pile is a queue of unreviewed improvements waiting on a manager.
When work happens
A day-of-week by hour-of-day heatmap of run starts. It highlights your Peak window with the number of runs started, and every cell is tappable for the exact volume, with a Less-to-More legend. Use it to see when your teams really operate: it's the honest input for shift planning, coverage, and setting claim SLAs on externally triggered work.
5. The workflow-level Insights tab
Open any workflow and switch to its Insights tab for the zoomed-in view. Four headline counters sit at the top: Open flags, Force-skips, Worker notes, and Bottlenecks.
Step-by-step analytics
Average time per step, skip rate, and flag count, rolled up across every version of the workflow, so history survives republishing. Each row shows the step position, how many runs reached it, its skip percentage, and flag badges; healthy steps get a clean marker. Steps whose required status was bypassed carry the warning Required step bypassed by worker(s) plus a force-skip count badge and excerpts of the skip reasons workers gave. Until there are runs, the panel reads: no runs yet, analytics will populate as workers complete the workflow.
Force-skipped required steps
Every case where a worker bypassed a step the SOP says is required, each with the reason they typed at the moment of skipping. Treat these as your highest-grade evidence that the SOP and reality disagree: workers only force-skip when the required step is actually blocking them. Each row links to Open run so you can see the full context. Very old skips recorded before reasons were captured show a legacy note instead of a reason.
Flagged issues
Every flag a worker has raised against this workflow, with its type, message, and resolution state, including who resolved it. Unresolved flags are the ones still waiting on a manager.
Worker notes
The qualitative signal: everything workers wrote in step notes while running this workflow. Notes are the raw material the optimization engine mines, and they're often where the real story lives ("the label printer on dock 2 has been broken for a week").
6. From insight to improvement
Insights is the observation half of a loop; the optimization inbox is the action half. The nightly analysis reads the same signals you see on these pages (skips, force-skip reasons, flags, timings, worker notes) and turns them into concrete suggestions.
A practical weekly routine for managers:
- 1Open Insights on 7 days and scan the six KPI deltas for anything moving the wrong way.
- 2Check Step bottlenecks and Most-flagged steps for repeat offenders.
- 3Drill into the worst workflow's Insights tab and read the force-skip reasons and worker notes. The "why" is almost always written there.
- 4Review anything waiting in the optimization inbox, or press Optimize on the workflow to request a fresh analysis.