Service reports, dashboards and alerts
Until now the no-code report builder could only see records: cases, companies, contacts. Everything that lives in a satellite table (satisfaction surveys, calls, time logged on a case, support entitlement usage, AI reply suggestions, incidents, self-service sessions) was out of reach, so questions l
Build your own service report
Until now the no-code report builder could only see records: cases, companies, contacts. Everything that lives in a satellite table (satisfaction surveys, calls, time logged on a case, support entitlement usage, AI reply suggestions, incidents, self-service sessions) was out of reach, so questions like "satisfaction by category" or "hours worked per customer" needed someone to write code.
Eight service data sources are now in the builder, next to the ones you already used:
- **Service satisfaction (CSAT)** and **Service loyalty (NPS)**: scores, effort score, when the survey was sent and when it was answered, promoters and detractors. - **Service calls**: wait, talk time, after call work, outcome, voice queue and agent. - **Time logged on cases**: minutes per case, per person, billable or not. - **Support entitlement usage**: cases and minutes charged to each entitlement. - **AI reply suggestions**: how many were offered, how many were accepted, and on how many cases. - **Incidents**: severity, root cause, when it started and when it was resolved. - **Self-service sessions**: sessions that ended without a case and sessions that became one.
Open Reports, pick one of them, choose what to group by and what to measure. From there it is a report like any other: you can save it, share it, schedule it by email and export it.
Named dashboards you can hand to the team
Dashboards used to be personal. Now you can create a dashboard with a name, put blocks in it and share it with a person or a whole team, exactly like a report. A block shows either a service metric or a report you saved.
Two things worth knowing:
- A metric block follows the dashboard period. A report block uses the period saved inside the report, and the block says so: we do not rewrite someone else's report behind their back. - A block with no basis in the period shows a dash, not a zero. "Nobody answered a survey" and "the score was zero" are opposite things.
Alerts about your own operation
Anomaly detection tells you something changed. An alert tells you something left what you agreed: "warn me when open cases go above 200", "when SLA compliance drops below 90%". Write the metric, the comparison and the limit, and pick how often it may repeat at most.
Use **Test now** before trusting it: it evaluates the alert and shows what would happen, without recording anything and without notifying anybody. A metric with no basis in the period never fires, and says so.
You can also turn on the **weekly service summary**, which sends the executive read of your service operation to the people who administer the account.
Grouping similar cases into an incident
When one outage produces a hundred cases, linking them one by one is not work, it is typing. Turn on grouping and the product shows, for each open case with no incident, which open incident the similarity points to and how many similar cases are already there. One click links it.
Automatic linking is opt-in and off by default, and there is one honest limit: if your account has no embeddings, the similarity comes from text search, whose scale is different. In that case the product only suggests and tells you why, instead of promising an automation that would never fire.
Every decision is recorded, including the ones where nothing was linked and why. Resolving an incident does not close the linked cases on its own: that is still an explicit action, because closing a customer's case is a decision for the person handling it.