Workspaces & Configuration

A workspace is the unit that owns calls, users, goals and QA rules. Get its configuration right and every number downstream means what you think it means.

What a workspace is

A workspace holds:

its own calls — uploads and connected-source scans target one workspace;
its own users — people are invited to specific workspaces;
its own goals — duration, calls per day, talk ratio, emotion mix, score weightage;
its own QA rules — masking, transcription language, dead-air threshold, processing limits;
its own saved charts in Analytics.

The top-bar switcher decides which workspace the dashboard, analytics, performance and call screens are showing. Nothing crosses between workspaces, which is exactly why the boundaries are worth thinking about before you create the second one.

How to structure them

Split by	When it’s right	Watch out for
Process or campaign	The usual answer. Collections, sales and support need genuinely different scoring and different targets.	Nothing — this is the split the goals and weightage system is designed for.
Client or brand	BPOs serving several clients, where reporting and access must stay separate.	Configuration is duplicated per workspace; set a template mentally before you create ten.
Language	Where a process runs in materially different languages and you want the transcription setting to match.	Splitting fragments an agent’s numbers if the same agent works both.
Team or shift	Rarely worth it.	You lose floor-wide comparison; the agent selector and filters already answer this.

The workspaces list

Sidebar › Manage › Workspaces lists every workspace with uploaded calls, total users, created date and status, in either grid or list view, with a search box for large accounts. Create New is top right; clicking any workspace opens its detail screen.

The workspaces table with a row per workspace showing uploaded calls, total users, created date, status and a View Details action — The list itself. The workspace you are currently in is highlighted and marked **Current**; **View Details** at the end of a row opens it.

Creating a workspace

Name it Use the process name your team already says out loud — "Collections – Bucket 2" beats "WS2".

The dialog takes a name and a description. Everything else — goals, masking, language and dead air — is set on the workspace once it exists.
Describe it One line on what the process is, for whoever inherits it.
Create, then configure Open the new workspace and set goals, masking, language and dead air before you point calls at it.

The workspace screen

The detail screen has a summary rail on the left and the editable sections on the right.

The workspace detail rail: identity card with initial, name, workspace ID and an Edit Workspace button, above a summary card — The left rail. Identity on top, then the summary — uploaded calls, users, created date, language, masking state and dead-air threshold in one glance.

Identity Name, a Current Workspace tag if it's the one you're switched to, the workspace ID, and Edit Workspace for name and description.
Summary Uploaded calls, total users, created date, language, PII masking state and the dead-air threshold — the fast configuration check.
Users Assigned · Goals / Targets · Configuration The three editable sections, each with its own edit control.

Goals & targets

This section is what turns raw metrics into judgements. Everything here feeds the dashboard KPI rings, the benchmark strips on the charts, the duration warning on the call list, and the goal column in the agent league table.

Setting	What it drives	Guidance
Emotion targets happy · sad · neutral · fear · disgust · surprise	The benchmark comparison on emotion charts, per call and in aggregate.	Set as percentages of the mix. Start from your own last month rather than an ideal.
Call Opening / Handling / Closing weightage	How the three stage scores combine into the overall call score.	Weight the stage your process actually lives or dies on — verification for collections, pitch handling for sales.
Overall Sentiment	The target sentiment for a good call — positive or negative.	Positive for almost every process.
Calls Per Day	The per-agent volume target behind the Goal column and the dashboard’s calls ring.	Use your real roster expectation; the league table multiplies it by the days in range.
Average Duration (minutes)	The duration KPI ring, and the red duration warning on individual calls.	Set it slightly above your median handling time so only genuine outliers turn red.
Percent Talk	The talk-ratio benchmark on the call screen.	Most processes sit healthy in the 40–60% band; check yours before committing.
Cross Talk	The interruption benchmark.	Lower is better. Keep the target tight — cross talk is a strong escalation predictor.

The Goals / Targets card on the workspace screen, grouped into sections with an edit control in its header — Every target on one card — emotions, stage weightage, calls per day, duration, talk and cross talk — behind the edit control in its header.

Change weightage deliberately

Opening / handling / closing weights change what every call score means. Adjusting them mid-quarter makes before-and-after comparisons misleading. Pick them once, then leave them alone for the review period.

Configuration

Setting	Options	Effect
PII Masking	On / Off	Redacts personally identifiable details — names, account numbers, ID numbers — from transcripts. See Security & Data handling.
Primary Language	English / Other	The transcription language for the workspace. Other covers non-English processes, including Hindi and code-switched Hinglish conversations.
Dead Air Threshold	3–10 seconds	How long a silence must run before it counts as dead air in the silence metrics.

A single Edit configuration control updates masking, language and dead air together, so a configuration review is one pass rather than three.

The Configuration card listing PII masking, primary language and dead-air threshold with their current values — All three settings on one card, each with its current value — masking, language and the dead-air threshold.

Processing limits

Limits control which calls are worth spending minutes on:

Minimum Call Duration (minutes) — skips calls too short to contain a real conversation, such as dropped and misdialled calls.
Maximum Call Duration (minutes) — skips outliers, typically recordings left running.
Percentage of calls to process — samples a share of each agent’s calls rather than all of them.

On sampling

Sampling saves minutes but reintroduces the blind spot that full-coverage QA exists to remove — risk does not conveniently appear in the sampled share. Use it to control a pilot's cost, not as a steady state.

Users assigned

The Users Assigned card lists everyone who can see this workspace, and its edit control adds or removes them. Adding a user here grants access to this workspace only; creating the account itself happens under All Users. Admins reach every workspace regardless.

When changes take effect

Goals, targets and weightage apply to how scores are presented and compared, so dashboards and league tables reflect them straight away.
Language, masking, dead air, keywords and limits apply to calls processed after the change. Calls already processed keep the settings that were in force when they ran.
If you change configuration to fix a systemic problem, re-ingest a sample batch afterwards to confirm the new behaviour before you trust the trend.