The dashboard

🔍

NiroHelp → Dashboard is the landing page of the NiroHelp menu and the first
screen a support manager sees. It answers, in order: what happened, what do I do
now, and where is this heading.

Every panel reads your own site’s figures. Nothing on this screen waits on the
AI service, so a slow connection cannot stall it.

The date range #

One control in the header — 7 / 30 / 90 days — drives every panel that has
a window. Reset puts it back to the default.

Today’s briefing #

A written summary rather than another chart: how much came in, what is waiting,
how much the auto responder handled, and your biggest documentation gap.

It is generated once a day, at 6am your site time, not when you open the
page. The card says so, and shows when it was written, so nobody reads it as a
live number.

Two cases where it falls back to plain figures instead of prose:

• Too little activity to summarise. You get open, waiting, resolved and
  median first reply, labelled “from your figures”.
• You changed the date range to one the stored summary was not written for.
  Better than describing a 7-day window with 90-day prose.

The headline figures #

Eight tiles, each with its change against the previous window:

Tile	What it counts
Open tickets	Everything not resolved, closed or cancelled.
Waiting on us	Tickets where the ball is in your team’s court.
Median first reply	Time from a ticket arriving to the first agent reply.
Resolved	Resolved inside the window.
Docs published	Published docs in the library.
Doc helpfulness	The share of doc votes that were positive.
AI deflection	Chatbot conversations that never became a ticket.
AI replies sent	Auto-responder replies posted in the window.

Client sentiment #

A positive / negative / neutral split for the window, beside the briefing.

Tickets — three queues #

Triage, ordered so the first thing you look at is the first thing to do:

Queue	Ordering
Needs attention	By risk
Needs reply	Oldest first
New tickets	Last 24 hours

Each links through to the matching ticket list.

Recent activity #

A merged feed of ticket events, published docs and AI replies — raised, closed,
answered by the auto responder, now live in the knowledge base. Filter it by
All, Tickets, Docs or AI.

AI performance #

Tickets answered by AI lists the auto responder’s replies with their
confidence score and what happened next — No follow-up needed or Agent took
over. A high take-over rate is your signal to raise the confidence
threshold, and the card links straight to that setting.

Chatbot responses shows how many visitors asked, how many were resolved
without a ticket, how many escalated into one and how many walked away
mid-conversation, plus the top questions.

Docs — four panels #

Panel	Shows
Most read	Docs by view count. Previews and views by people who can edit the doc are not counted.
Gaps	Searches that returned nothing, busiest first — the docs you have not written yet.
Rated down	Docs readers voted down.
Topics	Docs per topic, so an empty chapter is visible.

Gaps is the panel worth acting on. Every time someone searches your docs and
finds nothing, the phrase is recorded. It is your readers telling you what to
write next.

Agents #

Per-agent load — open against their cap, an at-capacity badge, median first
reply and resolve times, and how much of their volume the AI helped with.
Unassigned tickets get their own row so nothing hides. Manage agents goes to
NiroHelp → Agents.

Hot spots #

Three heatmaps by hour of day over the selected window: tickets, agent replies
and chatbot conversations. Each labels its own peak. The week starts on
whichever day your Settings → General → Week Starts On says.

Ask NiroHelp #

Ask in the header opens the admin assistant, which reads these same figures
conversationally. See Ask NiroHelp.

Empty panels versus errors #

Panels say so in words when they have nothing to show — “No topics exist
yet.”, “No doc has been voted down yet.” An empty panel means no data, not
a broken screen.

The exceptions are the AI usage meter and the two AI log screens, which report
an error. Those read data the AI service does not offer yet.