Using the app
Call metrics
Booking rate and the other call numbers on the Insights Scorecard: what each one means and what to do when it moves the wrong way.
The call metrics answer three questions: are calls turning into tours, is the assistant fast and accurate, and are your communities picking up when they are needed. They are on the Insights Scorecard, in its Calls and Booking pages parts. They used to be a separate Analytics page; the old address opens the Calls part.
They come with every plan: you do not need the Conversation Intelligence add-on. They count every conversation, whether or not it has been analysed.
Every number uses the filters at the top of Insights: Look at (the whole account, a community, a playbook, a version, or a playbook as one community runs it), the dates (each community's own days; the last 30 complete days unless you choose), Channel, Came in through and Callers. Test calls are excluded unless you switch Include test calls on.
Info
The numbers are measured one channel at a time, phone calls unless you choose another. A rate over phone calls and website chats together would describe neither. Choose Every channel and the Calls part shows the booking rate for each channel side by side.
What the words mean#
These are the same everywhere in the app, in Ask and in the API:
| Word | Means |
|---|---|
| Connected | A conversation someone actually had: the call was answered or ran for some time. |
| Booked | The conversation made at least one confirmed tour. Counted once per conversation. A later cancellation does not undo it. |
| Booking rate | Booked ÷ connected conversations. |
| Minutes | Connected time, not rounded. Your invoice rounds each call up to your plan's increment, so its minutes can be a little higher. |
| Answered | Only on Usage: the billing meter. It differs from connected, for example for spam calls. |
The tiles#
| Tile | What it counts | What to do about it |
|---|---|---|
| Booking rate (of connected conversations) | Conversations that booked a tour, out of those that connected. | The headline number. If it falls after a publish, compare versions with the filter. |
| Qualified booking rate (of callers who were a fit) | The same bookings, out of callers who were a fit. | A low number here with a healthy booking rate means your not-a-fit rules are too tight, or your booking rules ask for too much. |
| Transfer rate (of connected conversations) | Conversations handed to a person. | Rising means the assistant is hitting questions it cannot answer. Check the questions it could not answer. |
| Conversations | Every conversation, and how many connected. | Sanity check against your own call volume. |
| Response time, typical | The median time from the caller finishing to the assistant starting. | Under a second feels like a person. If it climbs, tell us. |
| Response time, slowest 5% | The same at the 95th percentile. | A big gap between this and the median means some calls feel laggy even when the average looks fine. |
| Cost per conversation | The estimated vendor cost of a conversation. | Useful context, not a bill. Your invoice comes from Usage. |
| Cost per tour | The estimated cost per conversation that booked a tour. | The number to watch. A longer call that books is cheaper than two short ones that do not. |
Info
Cost tiles are an estimate of what a call costs to run. They are not what you are charged. Your charges are on the Usage page and your invoices.
The charts#
Every chart has a Show as table toggle, so you can read the exact figures.
Booking rate by community#
Booked ÷ connected conversations at each community, with the counts behind each rate.
Where callers drop off#
The share of callers who reached each stage and hung up there: Greeting, Questions, Answering questions, Booking, Wrapping up, Transfer.
| If callers drop off at | Look at |
|---|---|
| Greeting | The greeting may be too long. Keep the disclosure to one sentence and the opening question short. |
| Questions | Too many questions before anything useful. Lower Questions before giving something useful on the line of questioning. |
| Answering questions | The assistant is not answering well. Check the questions it could not answer and add to Knowledge. |
| Booking | The times offered do not suit. Widen the community's booking windows, or check the minimum notice. |
How conversations ended#
Every outcome in the period. A healthy pattern has bookings and callbacks at the top and very few Booking failed or Something went wrong.
Why calls were transferred#
What made the assistant hand a caller to a person: the caller asked, a booking rule sent them, the assistant could not answer, the caller sounded upset, a guardrail was triggered, an emergency, or a system failure.
"The agent could not answer a question" rising is a content problem, and you fix it in Knowledge. "The caller asked for a person" rising is usually fine: some families simply prefer a person.
Questions the assistant could not answer#
Your backlog for Knowledge, and the single most useful list here.
- With the Conversation Intelligence add-on, the card links to Knowledge gaps in Insights. Knowledge gaps groups the questions, shows the communities affected, and tracks each fix.
- Without it, the card lists the Top unanswered questions as they were asked.
Handoffs by community#
| Column | What it means | What to do |
|---|---|---|
| Handoff rate | How often callers to this community needed a person. | Much higher than your other communities usually means its profile or knowledge is thin. |
| Main reasons | Why, with counts. | Tells you whether it is a content problem or a caller preference. |
| Transfers answered | How often somebody picked up. | A low number is a staffing signal, not an assistant problem. Check the backup number and the staffed hours. |
| Callbacks captured | How many callbacks the assistant took. | Compare with the transfer answer rate: a lot of callbacks usually means nobody is picking up. |
| Time to call back | The median time your team took. | Against your own target from the community's human backup plan. |
| Missed callbacks | Callbacks not returned in the target time. | Should be zero. These are families waiting. |
Follow-up pages#
Underneath the handoff table, when you send follow-up pages, is what they are doing.
The tiles are the funnel: how many callers said yes to a text, how many of those texts were delivered, how many pages were opened, and the typical time to first open.
Underneath, what callers did — watched the introduction, sent the page to someone, answered your questions, moved or cancelled the tour — measured against the pages that were opened. A page nobody opened tells you about your texts, not about your page.
Attended rate#
The number worth the most, and the one to read carefully.
It splits booked tours three ways: callers who opened their page, callers who got one and never opened it, and callers who never got one. If the page is doing its job, the first group attends more often than the second.
Warning
Attendance comes from your CRM, tour by tour. A tour nobody has marked up is counted as not known — never as a no-show. We will not infer that somebody attended just because they did not cancel.
That has a consequence worth planning for: if your counselors do not record whether a tour happened, this number will say so rather than guess. The panel tells you how many tours it could not account for. Marking tours up in your CRM is what turns this from a story into a measurement.
A tour that was cancelled is in none of the three groups. A cancellation is neither a show nor a no-show.
Booking pages#
The Scorecard's Booking pages part counts what your booking pages do, for a community and the dates at the top. The other filters do not apply to booking pages.
| Number | What it counts |
|---|---|
| Page views | Visits to a page, one per visitor a day. No cookies are used: a visitor cannot be followed from day to day. |
| Picked a time | Visitors who chose a time, whether or not they went on to book. |
| Still booked | Tours booked on the pages, or by the assistant through a page, that have not been cancelled. |
| Completion | Tours booked on the page and not cancelled, ÷ page views. |
| Cancelled | Tours cancelled by the person, the rep or an admin. |
| Moved | Tours rescheduled to another time. |
| Attended, No-shows | From your CRM after the day, as for calls. |
Bookings by source counts every tour booked in the range: on a Phone call, in the Web
chat, on a page's Booking page link, or on a page Embedded on a website. Page bookings by
UTM source splits page bookings by the utm_source their link carried, so add one to the links
you share. The tables give the same numbers for each page and each rep.
A weekly routine#
- Check Booking rate against last week, filtered to the same period length.
- Read the questions the assistant could not answer and add anything real to Knowledge.
- Look at Missed callbacks per community and chase them.
- Skim five transcripts from Where callers drop off's biggest stage.
- Compare the attended rate for callers who opened their page against those who did not.
When there is nothing to see#
No calls to measure yet means there is nothing in the range. Widen the dates, choose another channel, or call your test number and switch Include test calls on.