WhatsApp reports
How your WhatsApp is doing in four views —Service, Assistant, Team and Campaigns—: what each number measures, how many cases it is based on, what it is compared with and how to export it.
In WhatsApp › Reports you see how your assistant and your team handle conversations, and how your campaigns are doing. Every number comes with three things: its formula (the “How is it calculated?” button on the card), how many cases it is based on (“based on 48 closed conversations”) and how it compares with the previous period.
- Choose the view
Service, Assistant, Team or Campaigns (Campaigns shows up if your role can see campaigns).
- Choose the period
7, 30 or 90 days up to today, or Range to pick the days on the calendar (up to 92 days, and no earlier than a year ago).
- Filter if you want
By number (if you have more than one) and by team. The team filter does not apply to Campaigns: a campaign does not belong to a team.
- Export
Export CSV downloads what you are looking at, with the same filters, in the language of your screen. It opens in Excel with accents intact.
Service
- Conversations opened by the person: the new conversations whose first message was written by the person (not a template or a message from you), out of all new conversations.
- Assistant first response and Team first response: the median time between the message the person opens the conversation with and the first reply, split by who answered first. Below is the p90: 9 out of 10 replies took that long or less. It is never an average: a one-day reply does not hide the rest.
- Unanswered: the conversations the person opened that nobody has answered yet. It is a snapshot of today, so it is not compared with the previous period.
- Time to close: the median time between the start of a conversation and its closing.
- Wrote back within 7 days: of the closed conversations, how often the same person wrote again the following week. So the 7 days are complete, it looks at closings from one week before the period.
- The New conversations chart shows, per day (or per week for periods longer than 31 days), the ones the person opened and the ones we opened.
Assistant
- Conversations the assistant wrote in, out of the conversations with any message in the period.
- Resolved without a person: of the closed conversations, how many had no message from your team. A conversation someone took and gave back without writing counts as the assistant’s.
- Handed to a person by the assistant, with the breakdown of why (asked for a person, complaint, payment received…) and how many the automatic rule handed over.
- Replies that did not go out, with their reason. When another message arrives while the assistant is writing, it answers everything together: that is not a failure and does not count.
- Replies using the knowledge base: how many used at least one document from your base.
Team
- Conversations the team wrote in and messages sent (with the average per conversation).
- Wait after the handoff: the median time between a conversation being assigned to the team and the first message from a person. Assignments still waiting for a reply are counted separately.
- Closed by a person and closings with a classification (of what a person closed, how much was classified).
- The By person table shows the same for each team member, and Classifications of the closings how many conversations each classification has (a conversation with two counts in both).
Campaigns
- Sent, Delivered, Read, Replied, Not delivered and Asked to stop receiving (within 72 hours of the send). Delivered and not delivered are measured against what was sent; read, replies and opt-outs, against what was delivered.
- People who turn off read receipts never count as read: that number always falls a little short. A reply is the only proof that there was a person on the other side.
- The funnel shows how much is lost at each step, the By campaign table the same numbers for each campaign (with a link to its detail) and Why they were not delivered, the reasons.
- A contact with retries counts once, in the period of its last attempt.
How to read the comparison
- The previous period is a window of the same length right before yours: if you look at the last 30 days, it is compared with the 30 before.
- The arrow goes up or down with the real number, and its color says whether that is good: for times, failures and opt-outs, going down is good. Volumes (how many conversations, how many messages) have no color: more is neither better nor worse.
- “Small sample”: a percentage calculated on fewer than 50 cases (in this period or the previous one) is still shown, but without color or arrow, with the change written next to it. With so few cases the difference does not tell anything yet.
- “No previous period to compare” appears when the data did not exist yet in the previous period, and “Measured since…” when it does not cover your whole period either.
- A “—” means there were no cases: it is never a made-up zero.
Why does the report not show costs or satisfaction?
The cost of the channel is not part of these reports, and the satisfaction survey does not exist yet. When there is satisfaction data, it will be added as another view.
Why does a number not match someone else’s?
It is almost always the scope: each person sees the report for the conversations they can see. Check the line above the numbers.
How often is it updated?
It is calculated when you open the view (the time appears next to the period). If you come back to the same view within a minute, you see the same calculation.