--- title: 4.1. Configuration slug: conversation-intelligence/41-configuration docTags: createdAt: 2026-10-21T07:29:46.313Z --- **Target audience:** Admins, QA leads, Supervisors (where their role allows) **Description:** The setup that makes CQA score the right calls the right way: scorecards (**Quality Profile**), reference documents (**Knowledge Base**), external ground truth (**Tool Integration**), routing (**Assignment Rule**), people (**User Management**), and alerts (**Alert & Notification**). Everything here lives under **Manage › Configuration** in the left navigation. Data setup — Metadata, Connector, Access API, and Job & Schedule — is under **Manage › Data Integration** and is covered in [chapter 5](https://docs.exotel.com/conversation-intelligence/5-manage-data-integration). Usage & Limit and Policy sit in the footer of the left navigation and are covered in [chapter 7](https://docs.exotel.com/conversation-intelligence/7-usage-policies-and-support). Each page below follows the same pattern: a **page map** (what the screen is and what is on it), a **feature hub** where rules or roles need explaining, and **tasks** with step-by-step instructions. Menu items appear only when your role and account allow them. See [2.4 Navigation map](https://docs.exotel.com/conversation-intelligence/2-getting-started#24-navigation-map). **Recommended setup order for a new account:** Metadata ([chapter 5](https://docs.exotel.com/conversation-intelligence/5-manage-data-integration)) → Quality Profile → Knowledge Base (optional) → Assignment Rule → User Management → Alert & Notification. *** # 4.1.1. Knowledge Base **Path:** Manage › Configuration › Knowledge Base · **Who:** Admin The Knowledge Base is your account-level library of product, compliance, and reference documents. You link documents to specific KPIs so the AI can use them as context when evaluating those KPIs. It improves evaluation context and coaching suggestions without changing your Quality Profile rubric. ## Feature hub ### How Knowledge Base documents are used All reference documents live in one place: the **Knowledge Base** (Manage › Configuration › Knowledge Base). Upload a document once, then link it to the KPIs that should use it. You can reuse one document across several Quality Profiles and see which KPIs use it. ### How it works 1. An Admin uploads documents to the Knowledge Base. 2. Each document is processed until its status is **Indexed**. 3. On a Quality Profile, you link one or more indexed documents to the KPIs that should use them. 4. For new analyses that use those KPIs, the AI can reference the linked documents during evaluation and cite them. ### Limits - Up to **10** documents per account. - Up to **10 MB** per file. - Formats: **PDF, DOCX, TXT, MD**. - **Resource Name** must be unique in the account. - A document linked to KPIs cannot be deleted. Unlink it from the Quality Profiles first. ## Page map | Region / control | What it does | | ------------------------------- | ---------------------------------------------------------------------------------------------------------- | | **Add Resource** | Opens **Add Resources** to upload a document. | | **Type** and **Status** filters | Narrow the list by file type or status. | | **Refresh data** | Reloads the list to show the latest processing status. | | Table columns | **Resource Name**, **Type**, **Status**, **Linked KPI's**, **Chunks**, **Size**, **Added On**, **Action**. | | Status | **Pending**, **In Progress**, **Indexed**, **Failed**. Only Indexed documents can be used on KPIs. | | **Linked KPI's** | Shows how many KPIs use the document. Click it to open **Linked KPIs**, a list of those KPIs. | | Action menu | **View**, **Edit**, **Download**, **Delete**. | | Row selection | Select several rows to download or delete them together. Linked documents are skipped on delete. | ## Tasks ### Upload a document 1. Go to **Manage › Configuration › Knowledge Base**. 2. Click **Add Resource**. 3. Enter a **Resource Name** (unique in your account) and an optional **Resource Description**. 4. Drag and drop the file, or click **Browse**. Use PDF, DOCX, TXT, or MD, up to 10 MB. 5. Click **Add**. 6. Wait for the status to change from **Pending** or **In Progress** to **Indexed**. Click **Refresh data** to update the list. If the status is **Failed**, check the file type and size, then upload again. ### Link documents to KPIs 1. Go to **Manage › Configuration › Quality Profile**. 2. Create a profile, or open an existing one with **Edit**. 3. Open the KPI that should use the document. 4. Under **Ground truth for this KPI › Knowledge Base**, select one or more **Indexed** documents. 5. Click **Save**. New interactions analyzed with that KPI can use the linked material as evaluation context. To link documents in bulk, use the **KB Documents** column in the Excel template ([Create a profile from Excel](https://docs.exotel.com/conversation-intelligence/4-manage-configuration#create-a-profile-from-excel)). To see which KPIs use a document, click its **Linked KPI's** count on the Knowledge Base page. ### Replace or rename a document 1. In the document's **Action** menu, click **Edit**. 2. Change the **Display Name** or **Description**, or choose a new file under **Replace file (optional)**. 3. Click **Save**. Uploading a new file re-processes (re-indexes) the document. ### Delete a document 1. Unlink the document from every KPI that uses it (check **Linked KPI's**). 2. In the **Action** menu, click **Delete** and confirm. This cannot be undone. *** # 4.1.2. Tool Integration & Second-Order KPI Scoring **Path:** Manage › Configuration › Tool Integration · **Who:** Admin Before scoring selected KPIs, CQA can call your REST systems of record (for example, an order management system), take only the fields bound to that KPI, and attach them as labeled external ground truth. The AI then verifies agent claims against real data rather than the transcript alone. This is **Second-Order KPI Scoring**. ## Page map Open the **REST APIs** tab for Second-Order KPI Scoring setup: these are the systems of record the evaluator can call while scoring. On the REST APIs tab you can **Search by name**, filter by status (**All**, **Enabled**, **Disabled**), and sort by **Recently added**, **Recently updated**, or **Name (A–Z)**. Each integration card shows its auth type, number of mappings and fields, last update, and a **Test** action. ## Feature hub ### How a tool call works at analysis time - CQA resolves the URL parameters from the interaction's metadata. - It makes one HTTP call per unique tool and URL for that analysis, shared across KPIs that need the same call. - Only the response fields you selected on the KPI are sent to the AI, never the whole response. - The fields are attached as labeled external-data blocks to the bound KPIs before scoring. ### Tool Integration vs Inject to AI REST API integrations are different from **Inject to AI** on Metadata ([5.2 Metadata](https://docs.exotel.com/conversation-intelligence/5-manage-data-integration#52-metadata)). Inject to AI adds short metadata values (truncated) as context for every analysis. Tool Integration fetches live, structured data from your system for specific KPIs only. ### Failure behavior If a required tool cannot be resolved or returns unusable data, analysis fails closed: CQA does not silently score without verification. | Reason | Typical cause | What to check | | ----------------------- | ------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------- | | `TOOL_METADATA_MISSING` | A mapped metadata key is absent or empty on the interaction. | Ingestion mapping; sample interaction metadata; the tool's URL parameter mapping. | | `TOOL_UNAVAILABLE` | Tool HTTP error (for example, 4xx), tool disabled or missing, or selected response fields missing from the payload. | Tool URL and credentials with the customer system; field paths in the catalog; network allowlists. | For more causes, see [Troubleshooting](https://docs.exotel.com/conversation-intelligence/8-troubleshooting). ## Tasks ### Register a REST API 1. Go to **Manage › Configuration › Tool Integration** and open the **REST APIs** tab. 2. Click **+ New Integration**. 3. **Endpoint:** - **Name:** Unique in your account. It is shown in the KPI tool picker and as the data label the AI sees. - **Method:** **GET**. - **API URL:** A URL template with `{placeholders}` for values that come from interaction metadata (for example, `https://oms.example.com/orders/{orderId}`). - **Auth Type:** **None**, **Bearer Token**, **API Key**, or **Basic Auth**, plus optional custom headers. 4. **Metadata → URL mapping:** Map every `{placeholder}` in the URL to an interaction metadata key. Keys come from your Metadata settings ([chapter 5](https://docs.exotel.com/conversation-intelligence/5-manage-data-integration)). 5. **Response fields:** For each field the AI may use, enter a **JSON path** (use dotted paths for nested keys, for example `shipping.carrier`) and a **Description** of what it means. Add at least one field. 6. Click **Test connection**. Enter a sample value for each metadata key, then click **Test**. The result shows whether the call worked, which catalog fields were found in the response, and what came back. Nothing is saved during the test. 7. Save the integration. It is now available to bind on Quality Profile KPIs. **Tips:** - Declare only fields you intend to select on KPIs. - Keep metadata key names aligned with what your ingestion and connectors actually send. - If you change the configuration after testing, test again before relying on it. ### Bind tools on a KPI (Use Tool) 1. Go to **Manage › Configuration › Quality Profile** and **Edit** the profile. 2. Open the KPI that needs external verification. 3. Under **Ground truth for this KPI**, turn on **Tool Integration**. 4. **Select a tool**, then **Select response fields** for it. Use **Add Another Tool** to bind more than one. 5. Every selected tool needs at least one field. Remove a tool or pick a field, otherwise you cannot save. 6. Click **Save**. If no integrations exist yet, the KPI shows **Set Up An Integration**. If all integrations are turned off, turn one on in Tool Integration first. *** # 4.1.3. Quality Profile **Path:** Manage › Configuration › Quality Profile · **Who:** Supervisor (view only), Admin A Quality Profile is the structured rubric (scorecard) used to evaluate interactions — for example "Sales Audit" or "Customer Service Review". Assignment Rules decide which calls are scored against which profile ([4.4](https://docs.exotel.com/conversation-intelligence/4-manage-configuration#44-assignment-rule)). ## Page map | Region / control | What it does | | ------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Search | **Search by Quality Profile Name**. | | **Create Profile** | Opens a menu with the ways to create a profile: **Create Quality Profile** (manual), **Create Quality Profile with AI** and **Create Quality Profile - Excel Upload**. See [Tasks](https://docs.exotel.com/conversation-intelligence/4-manage-configuration#tasks). | | Table columns | **Quality Profile**, **Created At**, **Created By**, **Score**, **Profile Status**, **Action**. | | Action menu | **Edit**, **Duplicate**, **Download** (as Excel), **Delete**. | If your plan's limit on the number of Quality Profiles is reached, a message asks you to delete an existing profile or upgrade the plan before you can create another. ## Feature hub ### Profile structure A profile has three levels: 1. **Category** — a high-level section (for example, "Opening & Closing"). 2. **Sub-category** — a focused group inside a category (for example, "Greeting"). 3. **KPI** — an individual question the AI answers for each interaction (for example, "Warm Welcome"). ### KPI settings | Setting | What it means | | ----------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | **KPI Title** | Short name shown in reports (for example, "Warm Welcome"). **Must be unique within the profile.** The check ignores case and leading/trailing spaces, so "Greeting" and " greeting " count as the same title. If a title is already used, the editor shows: "Another KPI in this quality profile already uses this title. Titles must be unique within a profile." | | **KPI Input Type** | **Yes/No** — binary choice; assign points to Yes (for example, 5) and No (for example, 0). **Selection** — multiple choice; define scenarios (for example, "Option A: Mentioned both tax and amount", "Option B: Mentioned only amount") and the AI picks the best match. **Rating** — a rating scale for subjective measures; you set the maximum score (up to 20). **Text** — a data extractor that pulls information from the conversation (for example, "What competitor products did the customer mention?"). | | **KPI Question** | The full, detailed prompt for the AI (for example, "Did the agent introduce themselves by name?"). | | **Criticality** | Makes a failed KPI zero out the score of its **Category**, **Sub-category**, or the whole **Profile**, regardless of other passing KPIs. | | **Allow Scoring** | Turns scoring on or off for the KPI. Scoring is turned off automatically for the Text input type. | | **Ground truth for this KPI** | Optional sources the AI checks the agent's claims against: **Knowledge Base** documents ([4.2](https://docs.exotel.com/conversation-intelligence/4-manage-configuration#42-knowledge-base)) and **Tool Integration** fields ([4.3](https://docs.exotel.com/conversation-intelligence/4-manage-configuration#43-tool-integration-second-order-kpi-scoring)). | Your plan sets a maximum number of KPIs per profile. When you reach it, the editor disables adding KPIs, an Excel upload with more KPIs is rejected, and an AI-generated profile is trimmed to the maximum. ### Topics (for Smart Assignment) Topics describe when a profile should apply (for example, "Insurance Enquiry", "EMI Collection"). Smart Assignment matches call content against these topics to choose profiles ([4.4](https://docs.exotel.com/conversation-intelligence/4-manage-configuration#44-assignment-rule)). - Up to **10 topics** per profile. Duplicate topics are not allowed. - **Generate Topics** suggests topics from the profile name and description (add a description first). - Topics are optional. Profiles used only with Rule-Based Assignment do not need them. ## Tasks ### Create a profile manually Use this for small or simple profiles. 1. Go to **Manage › Configuration › Quality Profile**. 2. Click **Create Profile**, then **Create Quality Profile**. 3. **Profile details:** Enter a **Name** (for example, "Inbound Support") and a **Description**. 4. **Topics (optional):** Type a topic and press Enter, or click **Generate Topics** and click a suggestion to add it. 5. **Build the structure:** Click **Add Category**, then inside the category click **Add Sub-category**. 6. **Add KPIs:** Inside a sub-category, click **Add KPI** and fill in the [KPI settings](https://docs.exotel.com/conversation-intelligence/4-manage-configuration#kpi-settings). Each KPI title must be unique within the profile. 7. Click **Save**. ### Create a profile from Excel Use this for large or complex profiles that you prefer to build offline. 1. Go to **Manage › Configuration › Quality Profile**. 2. Click **Create Profile**, then **Create Quality Profile - Excel Upload**. 3. Click **Sample Excel** to download the template. 4. Fill in the template: - **Category** and **Sub-category:** Your sections (for example, "Soft Skills" › "Empathy"). - **KPI:** A short identifier for reports (for example, "Politeness"). Must be unique within the profile. - **KPI Question:** The detailed instruction for the AI. - **KPI Type:** "Yes/No", "Selection", "Rating", or "Text". - **KPI Option Label:** The options displayed as outcomes. - **KPI Option Weightage:** For Yes/No, one row for "Yes" and one for "No", each with points (for example, Yes = 5, No = 0). For Selection, one row per option. For Rating, the range the rating applies to. - **Criticality:** "Category", "SubCategory", or "Profile", where applicable. - **KB Documents** (optional): To use Knowledge Base context on a KPI, list up to **3** document names from your Knowledge Base in **KB Documents**, separated by commas. The documents must already exist in [Knowledge Base](https://docs.exotel.com/conversation-intelligence/4-manage-configuration#42-knowledge-base). - **Tool Integration** (optional): To use Tool Integrations for additional context on a KPI, list the Tools created in **Tool Integrations**, separated by semi-colons. 5. Drag the filled file into the upload area (or browse to it). CQA validates the structure and creates the profile. 6. Open the profile to make final changes before using it. If the file has more KPIs than your plan allows per profile, the upload is rejected with the file's KPI count and your maximum. Reduce the number of KPIs and upload again. ### Create a profile with AI Use this when starting from scratch: the AI generates a complete profile from your business context and sample calls. 1. Go to **Manage › Configuration › Quality Profile**. 2. Click **Create Profile**, then **Create Quality Profile with AI**. 3. Complete the five wizard steps: 1. **Business Details** — Describe your business, industry, and customers (for example, banking, insurance, e-commerce support). 2. **Call Use-case** — Describe the calls to be evaluated (for example, "Inbound support calls aimed at product clarity and first-call resolution"). 3. **Evaluation Criteria** — Say what to focus on, including critical KPIs such as script adherence, compliance checks, or customer satisfaction. 4. **Upload Calls for Analysis** — Drag and drop sample call recordings. Audio only: **MP3** or **WAV**. Up to 50 files. 5. **Upload reference documents (optional)** — Upload SOP documents (PDF, TXT, DOCX) or paste a link to Google Docs or Microsoft 365. Skip if you have none. 4. Click **Generate**. Processing typically takes 1–2 minutes depending on the number of files. 5. Review the generated categories, sub-categories, and KPIs. Adjust questions, scoring types, and weightages, and add or remove KPIs as needed. If the AI generated more KPIs than your plan allows, only the first ones up to your maximum are kept. 6. Click **Save**. The profile is immediately available for Assignment Rules. ### Edit an existing profile 1. Go to **Manage › Configuration › Quality Profile**. 2. In the profile's **Action** menu, click **Edit**. 3. Change profile details, topics, categories, or KPIs. KPI titles must stay unique within the profile. 4. Click **Save**. This also applies to a profile created from Excel: after creation, you edit it in the app in the same way. (The **Download** action gives you an Excel copy of the profile; editing that file does not update the profile.) Changes apply to interactions analyzed after you save. Interactions already scored are not re-scored. ### Duplicate or delete a profile - **Duplicate:** In the **Action** menu, click **Duplicate** and confirm. A new copy is created with the same settings. - **Delete:** In the **Action** menu, click **Delete** and confirm. This cannot be undone. *** # 4.1.4. Assignment Rule **Path:** Manage › Configuration › Assignment Rule · **Who:** Supervisor (view only), Admin Assignment Rules automate quality assurance by linking incoming calls to the Quality Profiles that should score them. Instead of manually choosing calls to audit, you define rules once and CQA applies them to every new interaction. ## Page map | Region / control | What it does | | --------------------- | ---------------------------------------------------------------------------------------- | | **Create Assignment** | Opens the rule editor with two tabs: **Rule-Based Assignment** and **Smart Assignment**. | | Rule list | Each rule with its **Assignment Name**, profiles, and status. | | Row actions | Edit, activate or deactivate, and delete a rule. | ## Feature hub ### Assignment types | Type | When to use | How profiles are chosen | | ------------------------- | -------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Rule-Based Assignment** | Fixed routing by metadata. | You select one or more Quality Profiles. Matching calls are always analyzed with those profiles. | | **Smart Assignment** | Topic-based routing. | You select candidate Quality Profiles. CQA matches the call content to each profile's **Topics** and can score the call against every matching profile. | ### How rules are evaluated As recordings arrive in CQA: 1. CQA checks every active rule's filters against the interaction's metadata — not just the first match. 2. For each matching rule, the rule's analysis percentage decides whether the interaction is sampled. 3. CQA collects the Quality Profiles from all matching rules into a distinct set. 4. One analysis runs per unique Quality Profile in that set. | Situation | Example | Result | | --------------------------------------------------- | --------------------------------------------------- | --------------------------------------- | | One rule matches and assigns one profile. | Rule A assigns Quality Profile 1. | One analysis (Quality Profile 1). | | One rule matches and assigns multiple profiles. | Rule A assigns Quality Profiles 1 and 2. | Two analyses, one per profile. | | Multiple rules match and assign different profiles. | Rule A assigns Profile 1; Rule B assigns Profile 2. | Two analyses, one per distinct profile. | | Multiple rules match and assign the same profile. | Rules A and B both assign Profile 1. | One analysis (deduplicated). | **Configuration tips:** - Use overlapping rules when you intentionally want one interaction evaluated under several profiles (for example, a Compliance profile and a Coaching profile). - If you want only one analysis per interaction, design rules so the matches resolve to a single distinct profile. Each analysis counts toward your usage: one interaction scored by two profiles counts as two analyses, while the interaction itself counts once. See Usage & Limit in [chapter 7](https://docs.exotel.com/conversation-intelligence/7-usage-policies-and-support). ### Smart Assignment profile roles | Field | Role | | ------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | **Default Profiles** | Always applied when the rule matches, in addition to any topic matches. | | **AI Match Profiles** | Candidates matched to the call by topic. Each needs Topics on its Quality Profile (up to 10). | | **Confidence Threshold** | How strictly topics must match: **Strict**, **Balanced**, or **Broad** presets, or **Custom** (drag the threshold slider to set the value). Higher confidence means fewer, more certain matches. | | **Fallback Profile** | Used when no AI Match Profile meets the threshold. | In the rule editor, each selected profile is labeled **Default**, **AI match**, or **Fallback**. For how a call scored by several profiles looks to reviewers, see [3.2 Interaction view](https://docs.exotel.com/conversation-intelligence/3-work-operational-workflows#32-interaction-view). ## Tasks ### Create a Rule-Based Assignment 1. Go to **Manage › Configuration › Assignment Rule**. 2. Click **Create Assignment**. 3. Enter an **Assignment Name**. 4. Keep the **Rule-Based Assignment** tab selected. 5. **Select Quality Profile:** Choose one or more profiles (for example, "Sales Profile"). 6. **Add filters:** Build conditions with your Metadata fields. For example, IF Queue Name IS Support\_L1 AND Talk Time IS GREATER THAN 60 seconds — only Support\_L1 calls longer than a minute are scored against "Sales Profile". Only metadata fields with **Filter Option** turned on appear here ([chapter 5](https://docs.exotel.com/conversation-intelligence/5-manage-data-integration)). 7. Set the percentage of matching interactions to analyze. 8. Click **Create**. The rule applies to new matching calls. ### Create a Smart Assignment Use Smart Assignment when different conversations should map to different profiles based on what was discussed, not only on metadata. **Before you start:** Add [Topics](https://docs.exotel.com/conversation-intelligence/4-manage-configuration#topics-for-smart-assignment) to every Quality Profile you will use as an AI Match Profile. 1. Go to **Manage › Configuration › Assignment Rule**. 2. Click **Create Assignment**. 3. Enter an **Assignment Name**. 4. Select the **Smart Assignment** tab. 5. **Add filters (optional):** Narrow which calls are considered, using Metadata fields (for example, Queue Name IS Support\_L1). 6. **AI Match Profiles:** Select the candidate profiles to match by topic. 7. **Advance Settings (optional):** Set the **Confidence Threshold**, **Default Profiles**, and **Fallback Profile** (see [Smart Assignment profile roles](https://docs.exotel.com/conversation-intelligence/4-manage-configuration#smart-assignment-profile-roles)). 8. Set the percentage of matching interactions to analyze. 9. Click **Create**. If calls are not being analyzed, see [Troubleshooting](https://docs.exotel.com/conversation-intelligence/8-troubleshooting). *** # 4.1.5. Alert & Notification **Path:** Manage › Configuration › Alert & Notification (page title: **Alerts & Notifications**) · **Who:** Admin Alerts surface quality issues, governance activity, adoption gaps, and usage warnings through scheduled email digests, so Supervisors and Admins can act without logging in to check. ## Page map | Region / control | What it does | | ---------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------- | | **Enable Notifications** | Master switch for the account. When off, no alert emails are sent, whatever the rule settings. | | **Daily digest dispatch time** | When daily digests are sent, in your local time zone. | | **Agent Summary Insights dispatch time** | When Agent Summary Insights digests are sent. Allowed between 8:00 AM and 11:00 PM (account time zone); independent of the daily digest time. | | **Alert Rules** tab | All alert events grouped by category, with cadence, recipient role, and on/off toggle. | | **Alert History** tab | Log of sent notifications. | | **Save Changes** | Saves all changes on the page. | ## Feature hub ### How it works CQA evaluates active alert rules on each rule's cadence (daily, weekly, or monthly). When a rule's conditions are met — for example, a critical KPI fails or an agent has not logged in — CQA builds one consolidated digest email and sends it to the configured recipients. Multiple events of the same type are batched into one email per run. ### Alert events | Category | Event | What it reports | Default recipient | Configurable | | --------------------- | -------------------------------- | ------------------------------------------------------------------------------------------------------------------ | ----------------- | --------------------------------------------- | | Quality & Scoring | **Critical KPI Failure** | All critical KPI failures in the period (for example, a missing compliance disclosure). | Supervisor | Cadence, recipients | | Quality & Scoring | **Score Below Threshold** | Count of interactions whose final score is below a threshold (for example, under 60%). | Supervisor | Score threshold (%), cadence, recipients | | Governance | **Dispute Raised** | Count of new disputes on AI-scored KPIs. | Supervisor | Cadence, recipients | | Governance | **Dispute Resolved** | Accepted vs rejected dispute counts. Sent to the agent who raised the dispute. | Agent | Cadence | | Adoption & Engagement | **Non-Logging Users** | Agents or supervisors who have not logged in within a window (name, email, last login). | Supervisor | Inactivity window (days), cadence, recipients | | Adoption & Engagement | **Agent Inactivity Nudge** | A nudge email to each agent who has not logged in within a window. At most one per agent per cycle. | Agent (self) | Inactivity window (days), cadence | | Commercial & Limits | **Usage Approaching Plan Limit** | Usage has reached the warning threshold of the plan cap (current vs committed, absolute and %). | Admin | Usage warning threshold, cadence, recipients | | Commercial & Limits | **Max Analysis Cap Reached** | The account hit its analysis cap; no further interactions are analyzed until the limit resets or the plan changes. | Admin | Cadence, recipients | For how caps work, see Usage & Limit in [chapter 7](https://docs.exotel.com/conversation-intelligence/7-usage-policies-and-support). ### Recipients - **Recipient role:** The default role that receives the alert (Admin, Supervisor, or Agent). - **Recipient Resolution:** One or both of: - **Specific emails** — addresses you enter. - **Respective users of the role** — everyone who holds the selected role. New users in the role start receiving alerts automatically. - Events sent directly to agents (**Agent Inactivity Nudge**, **Dispute Resolved**) have a fixed recipient; **Specific emails** is not available for them. ### Agent Summary Insights digests The **Agent Summary Insights** section adds three email digests based on [AI Insight › Agent Summary](https://docs.exotel.com/conversation-intelligence/3-work-operational-workflows#33-ai-insight-agent-summary): | Digest | Who receives it | | ---------------- | -------------------------------------------------------------------------------------------------------- | | **Agent Digest** | Each agent, with personalized coaching from their own Agent Summary. Goes to the respective agents only. | | **Team Digest** | Supervisors, with team-level insights. | | **Org Digest** | Admins, with organization-level insights. | - **Cadence:** Each digest has its own **Cadence** setting — choose how often that digest is sent. Digests are sent at the **Agent Summary Insights dispatch time**. - **Prompt** (optional) customizes how the email is written; **Reset to default** restores the standard prompt. - Privacy: Supervisor and Admin digests never include agent IDs or names in AI-generated text. Agent digests cover only that agent. - Choose at least one recipient mode before turning on a digest. - If the section shows **PAUSED**, the feature is turned off for your account. Your settings are kept and resume when it is turned back on. ## Tasks ### Turn on notifications 1. Go to **Manage › Configuration › Alert & Notification**. 2. Turn on **Enable Notifications**. 3. Set the **Daily digest dispatch time** (for example, 09:00 AM) to match your team's working hours. 4. Click **Save Changes**. ### Configure an alert rule 1. On the **Alert Rules** tab, find the event (events are grouped by category). 2. Click the expand arrow on the event. 3. **Cadence:** Daily, Weekly, or Monthly. (**Critical KPI Failure** and **Score Below Threshold** offer Daily or Weekly only — no Monthly.) 4. **Recipient role:** For example, Supervisor, Agent, or Admin. 5. **Recipient Resolution:** Select **Specific emails** (and add addresses), **Respective users of the role**, or both. 6. **Threshold Configuration** (where applicable): Enter the score threshold (%) for Score Below Threshold, the inactivity window (days) for Non-Logging Users and Agent Inactivity Nudge, or the usage warning threshold. 7. Turn the event on (blue toggle). 8. Click **Save Changes** at the top right. ### Set up Agent Summary Insights digests 1. Set the **Agent Summary Insights dispatch time** (between 8:00 AM and 11:00 PM). 2. In the **Agent Summary Insights** section, expand **Agent Digest**, **Team Digest**, or **Org Digest**. 3. Set the **Cadence**, choose recipient modes and, optionally, edit the **Prompt**. 4. Turn the digest on and click **Save Changes**. ### Review alert history 1. Open the **Alert History** tab. 2. Filter by **Event type**, **Time range** (All time, Last 24 hours, Last 7 days, Last 30 days), or **Status**. 3. Click a row's **Recipients** to see the individual addresses. | Column | Description | | ------------------- | ------------------------------------------------------------------------------------------------------- | | Timestamp | When the notification was sent. | | Event | The alert event type. | | Recipients | Who was targeted. | | Status | Delivery status (below). | | Notifications Count | Number of individual notifications in the digest (for example, 252 critical KPI failures in one email). | | Status | Meaning | | ------------------- | ----------------------------------------------------------------------------------------- | | Delivered | All intended recipients received the email. | | Failed | No recipients received the email. | | Partially Delivered | Some recipients received it and some did not (for example, a user with no email address). | *** # 4.1.6. User Management **Path:** Manage › Configuration › User Management · **Who:** Admin This is the canonical guide for inviting and managing users. [Chapter 1](https://docs.exotel.com/conversation-intelligence) gives only an overview of roles. ## Page map | Region / control | What it does | | ---------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | Role count chips | Number of users per role — **Admin**, **Supervisor**, **Agent**, and any custom roles — plus **Total**. Where your plan caps a role, the chip shows used vs allowed (for example, `8/10`). | | Search | **Search users...** by name, email, or username. | | **+ Create** | Opens **Create User** to invite someone. | | Table columns | **#**, **Name**, **Email Id**, **Username**, **Role**, **Status** (**Active** or **Inactive**). | | Access columns | One extra column per metadata field marked for **Access Control** (for example, Region), showing each user's value. | ## Feature hub ### Roles | Role | Typical access | | -------------- | ----------------------------------------------------------------------------------------------------------------- | | **Admin** | All Configuration and Data Integration pages, plus Usage & Limit and Policy in the footer of the left navigation. | | **Supervisor** | Work pages, Test Evaluation, Quality Profile and Assignment Rule (view only); resolves disputes and edits scores. | | **Agent** | Own scores, insights, Leaderboard, Report History; raises disputes. | | Custom roles | Your account may have additional roles (for example, QA). Their access is set up for your account by Exotel. | See [2.1 Choose your path](https://docs.exotel.com/conversation-intelligence/2-getting-started#21-choose-your-path) for what each role does day to day. ### Access metadata Access metadata are key/value attributes on a user (for example, `Region = North`). Combined with **Access Control** on a Metadata field ([chapter 5](https://docs.exotel.com/conversation-intelligence/5-manage-data-integration)), they restrict what data the user can see — for example, a Supervisor for Region A does not see Region B data. ### Usernames and invitations - Users sign in with their **username**, not their display name. If you leave **Username** blank, it defaults to the part of the email before `@`. Usernames cannot contain spaces. - The invited user receives a "Welcome to CQA!" email with a registration link. The invitation flow is described in [2.2.2 First-time login](https://docs.exotel.com/conversation-intelligence/2-getting-started#222-first-time-login-invited-users). ## Tasks ### Invite a user 1. Go to **Manage › Configuration › User Management**. 2. Click **+ Create**. The **Create User** drawer opens. 3. **User Profile:** Enter the user's **Email Id**. Optionally enter a **Username**; otherwise the part of the email before `@` is used. 4. **Access Settings:** Select the user's **Role** (for example, Agent or Supervisor). 5. **Access Metadata (optional):** Click **Add Metadata** and enter a key and value for each access attribute (for example, Region = North). 6. Click **Create User**. CQA immediately sends the invitation email. 7. Tell the user their username so they can sign in. If the user cannot accept the invitation (for example, the link has expired), see [Troubleshooting](https://docs.exotel.com/conversation-intelligence/8-troubleshooting). *** # 4.1.7. PII Redaction **Where:** No menu item — configured for your account by the Exotel delivery team · **Who:** Admin requests via CSM PII redaction protects sensitive customer data by automatically detecting and masking personally identifiable information in transcripts, scores, summaries, and reports. ## Feature hub ### How it works When PII redaction is configured, the analysis pipeline detects sensitive information in transcripts and masks it before the data appears anywhere in CQA. Redacted content is shown consistently in the transcript view, KPI result details, conversation summaries, and exported reports. ### Supported categories CQA can redact 11 categories: - Phone Number - Email Address - Person Name - Physical Address - PAN (India) - Aadhaar (India) - Bank Account - Credit Card - URL - IP Address - ID Card Number ### What to expect after enablement - **Transcripts:** Detected PII in enabled categories is replaced with placeholder text (for example, `[REDACTED]`). - **KPI results:** If a KPI answer references redacted PII, the redacted version is shown. - **Summaries:** Conversation summaries reflect the redacted content. - **Reports:** Downloaded and scheduled reports contain the redacted data, consistent with the app. Redaction applies only to interactions processed after it is configured. Previously analyzed interactions are not changed. ## Tasks ### Request PII redaction 1. Contact your Customer Success Manager. 2. Specify which PII categories to enable, or request all of them. 3. The delivery team configures your account. 4. Newly processed interactions are redacted according to your configuration.