AI Phone Genius
Docs/Analytics

Understanding Your Results

Who this is for: anyone on your team who wants to read the Analytics page, a call detail page, an Agent Scorecard, or a live Campaign Wallboard and actually understand what the numbers mean.

Your calling platform (branded "CallOps" in these docs — your provider may show their own name in the sidebar) turns every call into data automatically. You don't have to set anything up to see it. This guide explains what each number means, where to find it, and — most importantly — what to actually do when a number looks good or bad.

Where to find it

Page Sidebar link What it's for
Analytics Analytics Your main results dashboard — totals, trends, charts, agent leaderboard
Activity Activity A live, scrolling feed of everything happening across your account
Calls Calls Every individual call, with filters, drilling into full call detail
Agent Scorecard Open an agent → Scorecard button One agent's lifetime quality report
Campaign Wallboard Open a running campaign → Wallboard button A big-screen live view of one campaign in progress

The Analytics page — top numbers

At the top of Analytics you'll see seven number tiles. Here's exactly what each one means:

Tile What it means How it's calculated
Calls Total call attempts in the date range Every call record, regardless of outcome
Connect rate Percentage of calls where someone (or a voicemail system) actually picked up Calls with status "completed" ÷ total calls
Conversion Percentage of connected calls that ended in a booking Calls with outcome "booked" ÷ connected calls
Booked Raw count of calls marked "booked" —
Cost / booking Your average spend per booked meeting Total spend ÷ number of bookings
Spend Total wallet spend in the date range Sum of billed amounts on every call
Avg call Average length of calls that actually connected Average duration across calls with a duration greater than zero

A few honest notes on how these are computed:

  • Connect rate is about the call status, not the outcome. A call is "connected" if it reached status completed — meaning the call ran and ended normally (someone or a voicemail box answered and the call proceeded). It does not require a human to have spoken.
  • Conversion is deliberately booked ÷ connected, not booked ÷ total calls. This isolates your script/agent's closing ability from your list quality — a list of bad numbers will tank your connect rate but shouldn't be blamed on your agent's ability to book once someone answers.
  • Cost / booking and Conversion show a dash (—) when there's no denominator yet (no connected calls, or no bookings) — that's not a zero, it's "not enough data."
  • Avg call only counts calls with a nonzero duration, so it isn't dragged down by instant no-answers.

Date ranges

Use the range switcher at the top-right of Analytics to view 7 days, 30 days, 90 days, or 1 year. Every tile, chart, and the agent leaderboard update to that window. The default when you land on the page is 30 days. Ranges are calculated in whole days ending right now — so "7 days" is the last 7×24 hours, not the last 7 calendar days on a Monday-start week.

The charts

Call volume & bookings (left, large) — an area chart with two overlaid lines: total calls per day and bookings per day, across your selected range. Use this to see whether volume is climbing and whether bookings are keeping pace with volume, or falling behind it.

Outcome mix (right) — a donut chart of every call's outcome (booked, interested, callback, not_interested, no_answer, voicemail, opted_out, transferred, other), with a count badge under each. This is your fastest gut-check on where calls are landing overall.

Calls by hour of day — a bar chart showing what hour of the day (in your server's local time) your calls happen. Use this to find your busiest calling windows and to spot if you're dialing outside the times when people tend to pick up.

Calls by weekday — a horizontal bar list, Sunday through Saturday, showing call volume by day of week. Useful for noticing if certain weekdays are dead weight.

Agent leaderboard — a table ranking every agent that took at least one call in the selected window, sorted by call volume, showing calls, connect rate, bookings, and spend per agent. Each agent name links to that agent's full Scorecard (see below).

There is no separate "sentiment" chart on the Analytics page itself — sentiment breakdowns live on each Agent's Scorecard (see below).

CSV exports

Two exports are available directly from the Analytics page toolbar, plus two more elsewhere in the app:

Export Where to get it What's in it
Calls CSV Analytics page → "Calls CSV" button Every call in the selected date range: timestamp, direction, status, outcome, sentiment, duration, cost, contact name/phone, agent, campaign, and summary
Leaderboard CSV Analytics page → "Leaderboard CSV" button The agent leaderboard table for the selected date range: agent, calls, booked, connect rate, spend
Contacts CSV Contacts page → "Export" button Every contact in your account: name, phone, email, consent status, consent source, timezone, source, tags, date added
Campaign contacts CSV Open a campaign → "Contacts CSV" button Every contact in that one campaign: name, phone, email, dialing status, attempt count, last outcome, exclusion reason, next attempt time, date added

Exports respect the date range you had selected (for Calls CSV and Leaderboard CSV) and always respect your account — you'll only ever see your own data.

Reading a call's full detail

Click any row on the Calls page (or any date link from a chart drill-down) to open the full call record.

Recording & summary. If the call was carried on a real phone connection, a "Play recording" button plays the audio. Calls made in test/simulation mode don't produce audio — the button is grayed out and tells you why (no recording exists for mock or simulated calls). Below the recording button is a one-paragraph AI-written summary of the call, when one exists.

Transcript. The full text transcript of the conversation, turn by turn.

AI analysis. If the agent has post-call analysis fields configured, this card shows each extracted field (for example: "Interested?", "Best callback time," or whatever custom fields the agent was set up to capture) with its value pulled straight from that call. If a field was later removed from the agent's configuration, older calls still show it under its original name.

Outcome. A dropdown lets you (re)classify how the call ended: booked, interested, callback, not_interested, no_answer, voicemail, opted_out, transferred, or other. This is the same value that feeds every chart and export above, so keeping it accurate matters.

Important: setting a call's outcome to "opted out" is not just a label. The moment you save that outcome, the platform automatically:

  • Removes that contact's phone number from all active and future campaigns
  • Adds the number to your internal Do-Not-Call list
  • Logs the change as a compliance event

This is irreversible from the UI in the sense that it immediately blocks all future outbound dialing to that number — treat "opted out" as a real request to stop calling, not a casual status.

Details panel. Shows the linked contact, agent, campaign, the phone number the call went out on (or came in on), the underlying provider and its call ID, and the exact started/ended timestamps.

Sentiment. Where available, a small badge next to the call status (Positive / Neutral / Negative) shows the AI's read on how the conversation went overall. Not every call has a sentiment reading — only calls that went through post-call analysis will show one.

Agent Scorecard

Open any agent and click Scorecard to see that agent's entire lifetime performance, independent of date range:

  • Total calls, Booked, Avg handle time, Cost / booked — the same style of KPI tiles as Analytics, but scoped to this one agent across all time.
  • Sentiment mix — a set of horizontal bars showing the share of this agent's analyzed calls that came back Positive, Neutral, or Negative. Only calls with a sentiment reading count toward this.
  • Outcome breakdown — the same donut-chart treatment as Analytics' outcome mix, scoped to this agent.
  • 14-day call trend — a simple daily call-count chart for the last two weeks, so you can see if this agent's volume is trending up or down recently.

If the agent hasn't taken any calls yet, the Scorecard shows a friendly empty state pointing you to run a test call or launch a campaign.

Campaign Wallboard

Open a campaign and click Wallboard for a big, live, auto-refreshing view meant for a shared screen during a live calling session. It updates itself roughly every two seconds without you touching anything:

  • A status pill at the top ("Dialing live" with a pulsing dot when the campaign is actively running, or the plain campaign status otherwise) plus total contact count.
  • Four big counters: Active (currently calling or queued), Completed, Booked, and Opted-out / failed.
  • A donut chart breaking down every contact's current dialing status (pending, queued, calling, completed, failed, max attempts reached, excluded).
  • A "Latest outcomes" list showing the 10 most recently updated contacts and what happened to them, newest first.

This is a "watch it happen" view, not a historical report — it always reflects the current moment for that one campaign.

The Activity feed

The Activity link in the sidebar is a live, chronological feed of everything happening on your account: calls completed and booked, contacts imported or opted out, campaigns launched or paused, wallet top-ups, plan changes, and more. It polls for new events automatically and animates new rows in as they arrive. Use it as your "what just happened" screen rather than a reporting tool — it doesn't summarize or total anything, it just lists events in order with a human-readable one-line summary of each.

Practical reading guidance

What's a good connect rate? There's no universal number — it depends heavily on your list quality, calling hours, and area codes — but as a rule of thumb, a connect rate that's trending flat or dropping over the same list is a sign of a stale or low-quality contact list, not a script problem. If your connect rate is fine but conversion (booked ÷ connected) is weak, that's the opposite signal: people are answering, so the problem is what the agent says once they do.

When to change the script vs. when to change the list:

  • Connect rate is low or falling → look at your list, not your script. Check the "Calls by hour of day" and "Calls by weekday" charts to see if you're dialing at bad times for your contacts' timezone. Check whether the list is aging (stale numbers, wrong numbers, cell vs. landline mix) rather than tweaking what the agent says — a great script can't fix a call that never gets picked up.
  • Connect rate is healthy but conversion is low → this is a script/agent problem. People are answering and the agent isn't closing them. Compare the Outcome mix — if you're seeing a lot of "not_interested" or "callback" instead of "booked," revisit the agent's opening line and post-call analysis fields to understand where in the call people are dropping off (the transcript on individual calls is the fastest way to spot a pattern).
  • Cost per booking is climbing → check both. Rising spend with flat bookings usually means either the connect rate dropped (dialing more to get the same number of conversations) or conversion dropped (more conversations, same or fewer bookings). The Analytics KPI row makes it easy to see which one moved.
  • One agent's numbers are much worse than others on the same list → that's a script problem specific to that agent — check its Scorecard's sentiment mix and outcome breakdown side-by-side against a better-performing agent working the same campaign.

Related