The simple rule
One record = one feedback item, regardless of how much evidence Bagel AI finds.
A call recording with 12 product pains still counts as one item. A support ticket with zero mentions of a product gap still counts as one item. Bagel AI reads everything inside each source and extracts evidence, but the item count always reflects the number of items, not the number of findings.
What counts as one feedback item?
Source | % of items with gaps | Avg. gaps per item |
Call recording | 60% contain at least 1 gap | 2.3 gaps |
Survey | 45% contain at least 1 gap | 1.4 gaps |
Slack channel (per week) | 42% contain at least 1 gap | 1.6 gaps |
Support ticket | 15% contain at least 1 gap | 0–1 gaps |
CRM | Each CRM record is counted individually, including related objects such as loss reasons and health comments | — |
💡 The % column tells you how likely a given source surfaces a product gap at all.
💡 The avg. column tells you how many gaps to expect when it does.
💡 These are averages based on Bagel's data across sources.
Your numbers may vary depending on how your team uses each channel.
What does not count as a feedback item
Bagel AI operates entirely on customer feedback as its source of truth. Internal planning artifacts and business metrics are not treated as feedback.
The following are not counted as feedback items:
Epics or development tickets (Jira, Linear, etc.)
Roadmap/backlog items
Business impact metrics (ARR, LTV, revenue, etc.)
Product usage or analytics data
Feature requests are counted separately
Feature requests are tracked separately from feedback items.
Unlike feedback items, which Bagel AI discovers automatically by analyzing conversations, tickets, surveys, CRM records, and other customer signals, feature requests are submitted intentionally by a rep, CSM, support agent, or customer through a request channel.
Feature requests can be submitted through several channels, including:
Bagel AI's native Salesforce app
Bagel AI's native Zendesk app
The Bagel AI's Chrome extension
Customer-facing request channels such as Jira portals, intake forms, and other request submission systems
Because feature requests are manually declared, Bagel AI counts them as a separate object type rather than as feedback items.
That said, feature requests are not isolated from the rest of the customer signal. Bagel AI analyzes the content of each request, extracts supporting evidence, and connects it to related feedback discovered across your customer data sources. This allows you to understand both the volume of explicitly submitted requests and the broader customer feedback that supports them.
Supported Data Sources
Bagel AI can automatically discover feedback and customer signals from a wide range of systems.
Signal Type | Supported Sources |
CRM & Customer Records | Salesforce, HubSpot, Gainsight, Snowflake |
Call & Conversation Intelligence | Gong, Clari, Chorus, Momentum, |
Customer Support | Zendesk, Freshdesk, Intercom, Salesforce |
Surveys & Research | Qualtrics |
Community & Customer Discussions | Gainsight Community (Insided), Slack |
Product Feedback & Feature Management | Productboard, ProdPad, Notion, Coda, Airtable, Jira Product Discovery, Linear, Monday |
Project & Issue Tracking | Jira, Linear |
Data Warehouses | Snowflake |
💡Note: If your data lives outside one of our native integrations, you can connect it through our API or import data via CSV upload. This allows Bagel AI to analyze feedback and customer signals from virtually any system.
