Product tour

Every screen, and what each control does

Read this first, then open the demo and click everything. The demo is a real workspace with generated data — nothing you do there affects anyone, and it rebuilds itself every night.

▶ Open the live demo Start a free trial
1

Overview

Where you land. The health of the whole operation on one screen, for whatever period the filter bar is set to.

The four numbers across the top are the ones a contact centre is usually judged on. Each carries its movement against the previous period, so you can tell a bad day from a bad trend.

yourcompany.contactcurve.ai
Overview
👤 Agent
📈 Forecasting
👤 Attendance
🔧 Outbound Troubleshooter
Total calls
12,847
▲ 8.2%
Abandon rate
4.1%
▼ 1.3 pts
Avg handle time
6.4m
▼ 22s
Service level
87%
▲ 3 pts
1
Abandon rate counts calls that hung up before reaching anyone. It only means something if abandoned calls are actually in your data — ours keeps them, which many exports quietly drop.
2
Service level is the share answered inside your target. Set the target in the filter bar; the number recalculates as you change it.
3
The hourly bars are your day's shape. This is the picture the forecasting tab turns into a staffing plan.
2

The filter bar

Sits above every tab and drives all of them at once. Change it here and the Agent, Attendance and Forecasting views follow.

Changes are staged rather than applied as you type — pick an agent, a date range and a time window, then commit them in one go.

Filter bar
Agent: All ▾ Campaign: All ▾ Disposition: All ▾ 1 Aug — 31 Aug Time: 22:00 → 06:00 Today Last 7 days Apply
1
The Apply button turns amber when something is waiting. If your changes have not taken effect, that is what it is telling you — the bar is grey when the screen matches the filters.
2
Date presets act immediately. Asking to see last week is a request, not a field you are still editing.
3
The time-of-day filter matches the clock time a call arrived, not its date — so "what comes in overnight" is one question across the whole range. Setting From later than To is read as crossing midnight, which is the case it exists for.
3

Agent

Per-person handle time, after-call work, transfers and dispositions. Sortable, so the outliers surface without hunting.

Supervisors see only the campaigns assigned to them. An ops manager decides who covers what, so a team lead is not browsing another department's numbers.

Agent
Overview
👤 Agent
📈 Forecasting
👤 Attendance
Rita Alvarez
5m 12s
AHT ▼ 18s
Tom Bexley
7m 44s
AHT ▲ 51s
Priya Raman
4m 58s
AHT ▼ 6s
1
Handle time is talk plus hold plus after-call work — the whole time a call occupies someone, not just the part they were speaking.
2
A rising AHT is not automatically bad. Cross-check it against transfer rate: handling more yourself takes longer and is usually the better outcome.
4

Forecasting

Erlang-C staffing built from your own interval volumes, half hour by half hour. It answers "how many people do I need at 2pm on a Tuesday" rather than "how busy were we".

It reads history already collected and never calls Five9, so it is instant and cannot fail on a report limit.

Forecasting
Overview
👤 Agent
📈 Forecasting
👤 Attendance
Weeks of history: 12 Load history ● 11.4 weeks collected — good depth
1
The coverage line is amber below eight weeks and green above. Below eight the model is thin: it has not seen enough of your week to know what a normal Tuesday looks like.
2
History builds on its own once the sync is running. There is no backfill button, because a multi-week request to Five9 silently truncates — and a forecast that is quietly short is worse than none.
5

Attendance

Login and logout patterns by day, week and month, from the same Agent Login-Logout report your Five9 already produces.

One row is one shift, not one day — so an agent who logs in four times is four sessions, which is what makes total logged-in time and break patterns correct.

Attendance
Overview
👤 Agent
📈 Forecasting
👤 Attendance
Sessions
1,204
Avg logged in
7h 38m
Occupancy
72%
▲ 4 pts
Late starts
18
▲ 5
1
Occupancy is time on calls against time logged in. Sustained above about 85% is where burnout and attrition start — it is a warning light, not a target.
6

Getting your data in

Two ways, and they produce identical numbers.

On a paid plan, connect your Five9 account once and it collects on a schedule. On the trial, export the same two reports from Five9 and drop the CSVs in.

Workspace settings → Data
Import call log
Choose file — Call Log export (.csv)
Import attendance
Choose file — Agent Login-Logout export (.csv)
1
In Five9: Reports → Call Log Reports → Call Log, set your date range, run it, then export as CSV. Attendance is Agent Reports → Agent Login-Logout.
2
Five9 caps a report at 50,000 records. If you are exporting more than about a week of a busy queue, pull it in chunks — the import deduplicates, so overlapping ranges are safe.
3
Import tells you parsed, new and skipped. A file that is mostly "already stored" is a successful re-import, not a failure.

Now go and click things

The demo is a real workspace with generated data. Break whatever you like — it rebuilds itself every night at 3am.

▶ Open the live demo Start a free trial
No credit card · Nothing charged when the trial ends