Reporting & Analytics — Concepts and Glossary
Plain-language reference for every term used across the Reporting & Analytics guides. Each of the six audience guides links here rather than re-defining terms, so a definition lives in exactly one place.
If a term appears in a reporting guide and isn't defined here, it belongs here — flag it to your Connie contact.
Real-time vs. Historical
Two complementary views of the same underlying activity — driven by completely separate mechanisms on different clocks. Don't conflate their cadences.
- Real-time — what is happening right now, streamed live off Connie's event stream. There is no single fixed refresh interval; updates arrive as events occur. Answers: "What's going on this minute? Who's available, what's waiting?"
- Teams View — agent activity and status update as they happen (live stream).
- Queues Stats (Queues View) — queue-level metrics refresh about every 15 seconds; the "Now" tiles (active tasks, waiting tasks, available agents) refresh every 1–3 seconds.
- Historical — what has happened over time, and the trends inside it. Powered by Flex Insights, the reporting warehouse that accumulates every conversation. Answers: "What happened last month, and is it getting better or worse?"
The 60-minute figure you may see refers to the Historical (Adhoc) data pull — a separate mechanism from the real-time views. It is not how often Queues Stats or Teams View update. Real-time streams (≈15s for Queues metrics, 1–3s for the Now tiles, live for Teams); Historical refreshes on the adhoc cycle.
Basic vs. Connie Data Center (Advanced)
Connie reporting comes in two tiers.
The tier line is purpose + audience, not which tool builds the report.
- 🟦 Basic — operational reporting, included with every Connie account. For the people running the contact center (supervisors, admins). Answers "How is my center running?" — queues, SLA, handle time, agent activity. Native Flex Insights surfaced cleanly in the Connie UI, accessed from the left nav.
- 🟪 Connie Data Center (Advanced) — outcomes for external audiences; a paid upgrade. For funders, county, board, grants. Answers "What impact are we having, and can I hand it to a funder?" The connie.plus custom layer pulls your data out of Insights via API and reformats it into funder-ready deliverables. Native Insights produces operational dashboards; it does not produce funder-ready impact reports. Closing that gap is the entire reason the Connie Data Center exists. Accessed from the Connie Data Center destination; gated to upgraded accounts.
Within the Connie Data Center (Advanced) tier there are two kinds of work:
- Impact — the marquee use case (the what / why). Your CBO's effect on the communities it serves: closed-loop resolution, SDOH outcomes, community-referral coordination. This is what funders, boards, and counties want to see.
- Raw Data / Custom — the enabling capability (the how). Pull raw data via the Insights API and build/export anything (the connie.plus pipeline). The engine behind Impact.
The Flex Insights API is the seam between the tiers: Basic is "look at the dashboards"; Connie Data Center is "we pull your data and make it funder-ready."
A report's tier is set by who it's for and why, not by which tool produced it. A community-impact dashboard built inside native Insights today is still Connie Data Center (Advanced) — its purpose is an external-audience outcome. Don't let "it's built in Insights" pull Impact back into Basic.
The Drill-Down Model (KPI → Segment)
The defining capability of Connie reporting. Every report lets you move from the top-level number down to the single conversation behind it:
KPI (e.g. "Avg Handle Time: 4:32")
→ grouped by a dimension (e.g. by Department → PCA)
→ a list of conversations in that group
→ one conversation segment (talk, hold, wrap, transfer…)
Both out-of-the-box and custom views support this drill path. It's what turns a dashboard from a scoreboard into a diagnostic tool.
Metric Definitions
Connie's reporting is lifecycle-based: every task is tracked from arrival to completion and broken into segments, so the metrics below are all measurements of those segments.
Time metrics
| Metric | Definition | Note |
|---|---|---|
| Handling Time (Handle Time) | The time agents spend handling a task (including unavailable/offline activities tied to it). | Agent-side view. |
| Experience Time | The time a customer spends resolving their issue — including time in queue and time communicating with agents. | Customer-side view. |
| Talk Time | Time the agent and customer are actively communicating on the task. | A component of Handle Time. |
| Wrap Time (After-Task Work) | Time the agent spends finishing up after the conversation ends (case notes, etc.). | The "Wrap" lifecycle segment. |
| Hold Time | Time the customer is on hold during the task. | |
| Queue Time | Time a task waits in queue before an agent picks it up. | Part of Experience Time. |
These two will almost never reconcile 1:1 — Experience Time includes the wait, Handle Time doesn't. Read Experience Time for the true customer journey, then drill down to Handle Time to find improvement opportunities.
Volume & outcome metrics
| Metric | Definition |
|---|---|
| Handled | Conversations an agent accepted and worked. |
| Missed | Conversations offered to an agent that weren't answered. |
| Rejected | Conversations an agent actively declined. |
| Abandoned Conversations % | The ratio of abandoned conversations to total offered conversations. A customer left before reaching the team (e.g. hang-up, signal loss, timeout). |
| Abandon Time | How long a customer waited in queue before disconnecting. |
| AvgTalk | Average Talk Time across the group. |
| AvgWrap | Average Wrap Time across the group. |
| Quality % | Score from quality monitoring of handled conversations. In development — see note below. |
A voicemail or callback request is not counted as an abandon — these are tracked as Follow-ups.
As of June 1, 2026, Connie's Quality Dimensions — quality monitoring and reporting tooling — are in development. Customers will be notified when it's ready for beta testing. Until then, Quality % may appear blank.
Service Level (SLA)
Service Level (SLA) — the percentage of conversations answered within a target wait-time threshold. Two settings define it, set per queue and per channel (in seconds) by administrators under Service Level Preferences:
- SLA Threshold — the longest wait that still counts as "within SLA."
- Short Abandoned Time — waits too short to count against SLA, so a quick hang-up doesn't hurt the number.
The SLA shown in the real-time Queues Stats view is driven by Service Level Preferences. Connie's historical reporting has its own Service Level metric with separate threshold handling — the two are configured and calculated independently.
A queue showing 0% SLA usually means no threshold is configured yet, or every task ran past it. SLA targets are not set in Connie's config files (those only control which SLA tiles display and the Teams handle-time thresholds) — they're set in the admin UI. See Administrators → Basic Reporting to configure Service Levels.
Data-Model Dimensions
The axes you group and filter reports by. (Insights carries 235 attributes total; these are the ones that matter for Connie's use case.)
| Dimension | What it groups by |
|---|---|
| Department | The program a worker belongs to (e.g. Admin / H2H / PCA / RAMP). The cleanest grouping for program-level rollups. |
| Handling Department | The department that actually handled the conversation. |
| Agent Team / Handling Team | Team-level groupings. |
| Queue | The TaskQueue a conversation routed through. |
| Agent | The individual worker. |
| Communication Channel | Voice, SMS/Text, Webchat, Fax, etc. |
| Agent Location | Where the agent is based. |
| Agent Role | The agent's assigned role. |
Task Lifecycle (the segments behind the metrics)
Every conversation moves through these states; reporting segments map to them:
Pending → Reserved → Assigned → Wrap → Complete
- Pending — Connie is finding an available, qualified agent.
- Reserved — offered to an agent, awaiting accept/reject.
- Assigned — accepted; the agent owns it until they complete, park, reassign, escalate, or cancel it.
- Wrap — conversation done; agent finishing after-task work.
- Complete — closed.
An assigned task left untouched for two weeks (no complete/park/reassign/escalate/cancel) is auto-terminated by Connie and reported as an abandoned task.
Channels
Conversations are reported per channel and direction: Voice, SMS/Text, Webchat, Fax, across Inbound and Outbound.
How Connie's reporting categories map to the two tiers
Connie's underlying reporting breaks into four categories. Tier is assigned by purpose and audience, not by which tool builds the report:
| Connie category | Tier | Role |
|---|---|---|
| Real-Time (live dashboards) | 🟦 Basic | Operational — running the center |
| Historical (adhoc dashboards & reports) | 🟦 Basic | Operational — trends over time |
| Impact (community-impact dashboards) | 🟪 Connie Data Center | Marquee use case — outcomes for funders / board / county |
| Custom / Raw Data (export & reformat) | 🟪 Connie Data Center | Enabling capability — the connie.plus data pipeline |
Glossary index (A–Z)
Abandoned Conversations % · Abandon Time · Agent · Agent Location · Agent Role · Agent Team · AvgTalk · AvgWrap · Basic · Channel · Connie Data Center (Advanced) · Department · Drill-Down · Experience Time · Follow-up · Handled · Handling Department · Handling Time · Historical · Hold Time · Impact · KPI · Missed · Queue · Queue Time · Quality % · Real-time · Rejected · Service Level (SLA) · Short Abandoned Time · SLA Threshold · Talk Time · Task Lifecycle · Wrap Time
Draws from
- CACC User Guide → Data & Reporting — Handle vs Experience Time, Abandoned % + Follow-up exclusion, task lifecycle, four reporting categories, refresh/retention figures.
- Sprint record S11-260601 — the verified six standard metrics, Department (clean) vs Agent Team (messy), the two-tier model.
- Flex Insights API runbook — the 235-attribute data model and dimension list.
- Twilio Flex Insights docs — canonical Handling and Experience Time and data-model references.