Choose what you’re looking at
The filter row at the top scopes everything on the page:- Timeframe — four buttons: Day, Week, Month and Custom. Pick Day, Week or Month, then use the arrows beside them to step back through earlier periods; or pick Custom for a range of your own (up to one month), with a from and to date and time.
- Receptionist — one AI Receptionist, or All Receptionists for the whole workspace. Only the Receptionists you have now can be picked, so one you’ve removed isn’t listed — its conversations still count towards the All Receptionists figures.
- Timezone — the clock your figures are counted against. If your account offers more than one zone you can switch it, and your choice is remembered.
Analytics counts completed conversations and leaves out test chats. A call that’s still in progress joins the figures once it ends.
The tiles
The top of the page is a row of stat tiles. Each shows its figure for the period, and either how it compares with the previous period or what share of your conversations it accounts for. Where one direction is better, the comparison is green when the figure moved that way and red when it didn’t: more conversations is better, and fewer missed calls and fewer Transfers to Moneypenny are better. Average length and Transfers to You stay grey, because neither direction is better in itself.- Total conversations — every call and chat your AI Receptionists handled in the period.
- Avg conversation length — how long a conversation lasted on average.
- Missed calls — callers who tried to reach you but didn’t get through, shown as a share of everyone who tried.
- Transfers to Moneypenny — conversations your Receptionist handed to Moneypenny receptionists, shown as a share of your conversations.
- Transfers to You — conversations your Receptionist put through to your own people, teams or numbers, shown as a share of your conversations.
- Unanswered questions — conversations where a caller asked something your Receptionist couldn’t answer. If this is rising, it’s usually a sign there’s something worth adding to your Knowledge library.
- Spam and wrong numbers — conversations that weren’t from a real customer: automated calls and recorded messages, anyone selling to your business (sales, marketing, recruiters or surveys, whether or not they say so), scams, and wrong numbers. A genuine customer isn’t counted here, however short the conversation.
How callers felt
A bar splits your analysed conversations by how the caller came across — from Happy through Neutral to Frustrated. Each one is judged on how the caller felt by the end, whatever your Receptionist managed to do for them. Anything counted under Spam and wrong numbers counts as Neutral, whatever its tone, so it doesn’t skew the balance. Click a segment to open the Calls and messages page with that sentiment, period and Receptionist already applied — the natural next question after “a fifth of my callers left frustrated” is “which ones?”.Outcomes
Once your AI Receptionists have outcomes set up, this card says what the period’s conversations actually achieved. Each outcome gets a ring, and the list beside the rings names it, counts the conversations that reached it and gives that as a share. On a single Receptionist all ten of its outcomes get a ring; across all Receptionists the nine busiest are drawn. Anything beyond that is listed underneath instead, named and counted, under a line saying how many were too small to draw. Every share is of the conversations your Receptionists could read, and that figure sits in the middle of the rings. The shares happily add up to more than 100%, because one conversation can reach several outcomes: a call that booked an appointment and took a message counts towards both. Two things that sound alike are kept firmly apart:- None of your outcomes — the grey ring. These conversations were read and reached nothing on your list.
- Too short to judge — a line underneath, counting them. These were never read back at all, so rather than being counted as having achieved nothing, they’re left out of the shares and named separately.
- On a single Receptionist, every outcome on that Receptionist’s list keeps the same colour from one period to the next, so a quiet month doesn’t repaint the chart.
- On All Receptionists, outcomes with the same name are pooled across every AI Receptionist and the busiest are the ones drawn. The shares are of every conversation the workspace read, so an outcome only one Receptionist tracks reads lower here than on that Receptionist’s own view. The card says so, because the figure visibly changes when you switch.
Why callers didn’t get through
When any calls were missed in the period, a panel breaks them down by cause, so you can see whether callers hit busy lines or something you need to act on. It’s collapsed by default — the headline count is in the header.Conversations over time
The chart plots the period’s conversations — hour by hour for a day, day by day for a week, week by week for a month. Click a bar to zoom in: a week out of the month, then a day out of that week. Show data, at the top of the card, opens a table of the same figures with one row per bar. Weeks run Monday to Sunday, but a month rarely starts on a Monday or ends on a Sunday, so on the Month view the first and last bars usually cover only part of a week. Those are named for the days they actually take in — “1 – 6 Sept 2026”, or a single date where only one day of the week falls inside the month. The whole weeks between them are still named after the Monday they start on. Either way, only conversations inside the month you’re looking at are counted. Clicking one of those shorter bars opens the whole calendar week it belongs to, so “1 – 6 Sept 2026” takes you to the week beginning Monday 31 August. Expect a higher total there when the days either side were busy: that week’s view counts them, and the month’s figure doesn’t.Send report
Send report, at the top of the page beside its title, does two jobs from one dialog: it sends these figures now, or sets them to arrive regularly. Whichever you choose, the email carries the same thing the page does — the tiles, plus the How callers felt breakdown: the bar, a line beneath it naming each sentiment shown with its share, and how many conversations those shares are based on. If nothing in the period was analysed, that breakdown is left out.1
Select Send report
A line at the top of Send this report reads back what’s in it — the period, the Receptionist and the timezone — taken from the filters behind it, so the email can’t disagree with the page.
2
Say who it goes to
Send to starts with your own address. Use Add another for anyone else, up to ten in total; a recipient doesn’t need an account of their own.
3
Choose how often
How often starts on Just this once. Leave it there to send the figures on screen now, or pick Daily, Weekly or Monthly to have the report arrive on its own.
4
Send or schedule it
The button follows your choice: Send now, or Schedule it.
A recurring report always covers the last complete period, not the one on screen. Daily covers the day before, weekly covers the previous week from Monday to Sunday, and monthly covers the month before. The read-back in the dialog tells you which as you choose.