IVR reporting is three surfaces: a historical statistics screen with a live view, a place to define how calls are scored, and a per-call journey panel. Access follows call-analysis permission and agents never see any of it. The IVR entry only appears in the sidebar when the selected agency actually has a Router IVR queue.
The Three Surfaces
IVR Segment Statistics — historical performance plus a Live tab. Found under Calls, listed as IVR; the page breadcrumb reads Calls then IVR Segments.
Answer-time reports — where you define how calls are scored. Reached from IVR Segments. Needs an agency selected in the top-right picker.
IVR Journey — a read-only panel on a single call's detail page showing the exact path that caller took.
IVR Segment Statistics
The historical view: how many callers entered, how far they got, where they left, and how the answer-time target was met, organised by queue and date range. Alongside it is a breakdown by inbound number showing each number's friendly name where one is configured, which is how you tell whether one advertised number performs differently from another feeding the same flow.
The scorecard, the report snapshots and the Checkpoint detail panel sit on the Operations view of this screen, which is also where you find the stepped funnel from callers entering the flow through to the scored result.
⚠️ Note: Checkpoint detail and answer-time reports are not the same measurement. Checkpoint detail shows the raw timer rate — how often the timer was met, full stop. An answer-time report shows the scored result after your rules about abandons and exclusions. The two will disagree, and they are supposed to. If somebody quotes you a service level, establish which they are reading.
The Live Tab
The Live tab shows callers currently in the IVR, refreshing roughly every five seconds. It has two halves: callers waiting for an agent, and callers still moving through the flow before they reach a wait. Each caller is coloured by how long they have waited against your target.
⚠️ Note: Colour is measured from the Timer node, not from the start of the call. A caller who spent two minutes working through your menu and has only just reached the wait shows green, because their timer only just started. Say this out loud to anyone you sit in front of the screen — the intuitive reading is wrong.
Treat it as a floor view for catching calls approaching or breaching your target. It is a frequently refreshed snapshot rather than a real-time wallboard, and there are no supervision controls on it — you cannot listen, whisper or barge from this screen. Those controls live elsewhere in the dialer.
Answer-time Reports
An answer-time report is a saved definition of how calls should be scored: who is in the total, what counts as a success, and how to treat callers who hang up early. You can hold several, which is how one flow can be reported against both an internal target and a contractual one.
Name and description — when creating, quick-start templates prefill the rules: an 80/20 service level, a strict variant where abandons fail, or answer within 60 seconds at 80 percent. You still choose the timer.
Step 1, What you're measuring — the Timer Start span; who counts in the total; what counts as a success; which outcomes count as handled in time; and your goal percentage.
Step 2, Early hang-ups — exclude them, count them as failed, or count them as met. Plus a grace period, and the option to remove voicemail and IVR self-service from the counts entirely.
Step 3, Limit to phone lines — optional. Note that a queue filter applied in analytics overrides this.
Step 4, Default date range — optional.
Save — validates that you have a timer span, at least one handled-in-time outcome, a real abandon policy, any checkpoints your scoring depends on, and a name.
The choices that move your number most
What counts as handled. You can treat an agent pickup, a connection through to a queue, the IVR resolving the caller without an agent, and reaching voicemail as successes, in any combination — a win is recorded if any of the ones you tick match. These options disappear if you set success to reaching a checkpoint alone. Counting voicemail as handled is defensible for a service line and indefensible for a sales line.
How you treat early hang-ups. Excluding them removes them from the denominator and flatters the result. Counting them as failed is the strict reading and is what most contractual targets assume. If you are reporting to a health plan, assume they mean the strict reading unless the contract says otherwise.
⚠️ Note: The seconds and the percentage live in different places. The answer-time threshold in seconds comes from the Timer node in your flow, so changing it means editing the IVR. The goal percentage is a field in the report.
⚠️ Note: Where abandon figures show a dash rather than a zero, no saved report defines the split. That is not the same as no abandoned calls. Reading a dash as zero makes a queue look considerably healthier than it is.
Reading a Single Call
When you need to know what happened to one caller rather than how the flow performs in aggregate, open the call and look at the IVR Journey panel on the call detail page. It shows a verdict, badges for the milestones reached, the path step by step with the outcome and duration of each, and a table of timed spans. It is read-only, and calls that never went through an IVR do not show it. Call statistics on the same page are shown to the millisecond, which matters when arguing about whether a call missed a target by a second.
A Practical Weekly Rhythm
Daily, during hours — glance at the Live tab at your busiest hour. Look for callers going amber and red, not for a number.
Weekly — read the service level against your saved report and compare it to last week. Investigate the direction of travel, not the absolute figure.
Weekly — check the abandon split. If it shows dashes, fix the report definition before interpreting anything else.
Weekly — check the by-inbound-number breakdown for one number behaving unlike the others.
On any complaint — open the specific call and read the IVR Journey rather than reasoning from aggregates.
Monthly — re-read your report definition. Targets drift and the definition is easy to forget once saved.
Known Limits
The Live queue filter takes one queue at a time, so separate flows have to be watched separately.
No listening, whispering or barging from the Live tab.
Live is a frequent snapshot rather than a stream.
Live looks back about thirty minutes, so a caller waiting far longer can drop off the board while still on the line.
Colour runs from the Timer node, not from the start of the call.
A dash in the abandon split is not a zero.
The live view shows agent availability but not a full picture of who is currently on calls.