Zoom Contact Center Survey Best Practices
How to Identify Consumer Feedback on Live Agents vs. AI Bots
A Step-by-Step Configuration Guide for Separating What Your Customers Think About Your Bots from What They Think About Your People
As AI bots handle an increasing share of contact center interactions, one of the most critical operational questions is also one of the easiest to overlook: when a customer takes a post-interaction survey, do you know whether they’re rating a bot or a person?
This guide walks through the complete configuration process in Zoom Contact Center to set up surveys that cleanly distinguish between AI bot feedback and live agent feedback — from creating the surveys, to tagging engagement paths with flow variables, to building segmented reports in CX Analytics.
There are six phases. Each one builds on the previous. By the end, you’ll have a survey infrastructure that tells you exactly what your customers think about each handler type — not a blended score that masks both problems and wins.
Prerequisites: Zoom Contact Center license, Zoom Virtual Agent (ZVA) license (if using bot flows), admin access to the Zoom web portal, and at least one configured queue and flow.
PHASE 1
Create Separate Surveys for Bot and Agent Interactions
Build two distinct consumer engagement surveys — one with bot-specific language, one with agent-specific language — so responses are unambiguous from the start.
The single most common mistake is using one generic survey (“How was your experience?”) for all interactions. When a customer rates “the experience” a 2, you have no idea whether they’re frustrated with a bot that couldn’t understand them or an agent who put them on hold too long. Fix this at the source.
Step 1A: Create the Bot Survey
1) Navigate to Survey Management
Zoom Web Portal → Admin Center → Content management → Surveys → Consumer engagement survey tab
Click + Create and select Consumer engagement survey → Create.
2) Name and Configure the Bot Survey
Name it: Bot Interaction Survey - CSAT
Add your questions:
• Rating scale: “How satisfied were you with the virtual assistant’s help today?” (Bot-specific CSAT)
• Single choice: “Was your issue resolved by the virtual assistant?” Yes / No / Partially (Bot resolution tracking)
• Long answer: “What could the virtual assistant have done better?” (Open-ended bot feedback)
3) Customize and Save
Optionally customize the completion message, colors, and header image. Preview the survey to confirm the questions read correctly. Click Save to activate it, or save as a draft if you’re not ready to go live.
Step 1B: Create the Agent Survey
1) Repeat the Process for a Live Agent Survey
Follow the same navigation path and create a second survey named: Agent Interaction Survey - CSAT
Use agent-specific language:
• Rating scale: “How satisfied were you with the agent who assisted you today?” (Agent-specific CSAT)
• Single choice: “Was your issue fully resolved?” Yes / No / Partially (Agent resolution tracking)
• Long answer: “Is there anything else you’d like us to know?” (Open-ended agent feedback)
Why two surveys? Using distinct survey language is not just a UX nicety — it directly affects data validity. Gartner’s 2026 research found that customers increasingly cannot distinguish between bot and human responses. Explicitly telling them what they’re rating removes ambiguity from the data.
PHASE 2
Create Custom Variables to Tag the Engagement Path
Set up global custom variables that will record whether each interaction was handled by a bot, an agent, or both — before any survey fires.
Surveys tell you what the customer felt. Variables tell you what actually happened. You need both, and the variable should be set before the survey is triggered.
Step 2A: Create a Custom Variable Group
1) Navigate to Variables
Admin Center → CX Management → Contact Center → Preferences → Variables
Under Custom Variable Groups, click Add Custom Group.
Name it: Engagement Path Tracking
Click Add.
Step 2B: Create the interaction_type Variable
1) Add the interaction_type Variable
Select the Engagement Path Tracking group, then click Add Variable.
Configure it as follows:
• Variable Name: interaction_type
• Type: String
• Default Value: bot_only
• Description: Tracks whether engagement was bot_only, bot_then_agent, or agent_only
Click Save.
Step 2C: Enable Reporting and Analytics
1) Enable the Variable for Reporting
Still in Preferences → Variables, find the interaction_type variable in your group. Click the … menu → Edit.
Enable two separate settings:
• Use in reporting — Makes the variable appear in the Variables log (Contact Center Management → Logs → Variables)
• Use in CX Analytics and CX Insights — Makes the variable available as a filter/dimension in CX Analytics reports. Select the dataset group(s): Engagement and Queue
Click Save.
Common mistake: “Use in reporting” and “Use in CX Analytics” are separate settings. Enabling one does not enable the other. If you only check “Use in reporting,” the variable will appear in the Variables log but will NOT be available in CX Analytics reports. You need both.
PHASE 3
Configure the Flow to Tag and Route by Interaction Type
Wire up the flow editor so the interaction_type variable is set automatically based on whether the engagement stays with the bot or escalates to a live agent.
This is where the engineering happens. The flow editor is where you connect the variable to the actual customer journey.
Step 3A: Set the Default Value at Flow Entry
1) Add a Set Variables Widget at the Start of the Flow
Admin Center → CX Management → Contact Center → Flows → [Select your flow] → Edit
Drag a Set Variables widget onto the canvas and place it before the Bot widget.
Configure it:
• Variable: interaction_type (from Engagement Path Tracking group)
• Value: bot_only
This sets the default assumption: every engagement starts as bot_only. If the customer escalates to a human, we’ll update it downstream.
Step 3B: Update the Variable on Agent Escalation
1 Add a Second Set Variables Widget on the Transfer Exit
On the Bot widget, open Exits and find the Transfer to Staff Service exit.
Connect this exit to a new Set Variables widget (placed between the Bot and the Route To widget).
Configure it:
• Variable: interaction_type
• Value: bot_then_agent
Then connect this Set Variables widget to your existing Route To widget that sends the customer to the agent queue.
Step 3C: Handle Agent-Only Flows
1) Tag Agent-Only Interactions in Non-Bot Flows
If you have flows that route directly to an agent queue without passing through a bot (e.g., voice calls that skip the IVR bot), add a Set Variables widget at the start of those flows:
• Variable: interaction_type
• Value: agent_only
The resulting flow logic:
Flow entry → Set Variables (interaction_type = bot_only) → Bot widget
• Bot resolves → Survey widget (bot survey) → End
• Bot escalates via Transfer to Staff Service → Set Variables (interaction_type = bot_then_agent) → Route To (agent queue with agent survey linked) → End
The engagement ID stays consistent through transfers, so the interaction_type value follows the engagement from bot to agent to survey to reporting.
PHASE 4
Add the Bot Survey to Your ZVA Flow
Configure the Survey widget inside the Zoom Virtual Agent flow so bot-only interactions trigger the bot-specific survey.
Steps
1) Open the Bot Flow Editor
Zoom Web Portal → AI Studio → Virtual Agents → [Select your chatbot] → Bot Flows → Edit
If the flow is currently published, unpublish it first before making changes.
2) Add the Survey Widget
Drag the Survey widget onto the main (default) flow canvas.
In the widget settings:
• Add Survey: Select Bot Interaction Survey - CSAT (the survey you created in Phase 1A)
• Survey Prompt: “Thank you for chatting with us! We’d love your feedback on the virtual assistant.”
• Link Text: “Take a quick survey”
• Link Expiration: Set based on your preference (e.g., 24 hours)
3) Connect and Publish
Connect the Survey widget to the bot’s resolution exit — the path that fires when the bot successfully resolves the interaction without escalation.
Click Publish to make the flow live.
Important: The Survey widget should only fire on bot-resolution paths, NOT on the Transfer to Staff Service exit. If a customer escalates to an agent, the agent queue’s linked survey (configured in Phase 5) will handle the feedback collection. Triggering both would cause survey fatigue and data confusion.
PHASE 5
Link the Agent Survey to Your Queues
Attach the agent-specific survey to your live agent queues so it fires after human-handled interactions.
Steps
1) Navigate to the Queue
Admin Center → CX Management → Contact Center → Queues → [Select your agent queue]
Select the Survey tab, then click Add.
2) Select the Agent Survey and Configure Delivery
Choose Agent Interaction Survey - CSAT and click Save.
Configure delivery settings by channel:
• Voice: Immediate or delayed delivery; start/completion audio prompts; optional call-duration threshold
• Messaging: In-chat questions or survey link; survey prompt text
• Video: Survey link with configurable expiration; notification to agent/supervisor
3) Repeat for All Agent Queues
Repeat this process for every queue that handles live agent interactions. Each queue can have one linked survey, but the same survey can be used across multiple queues.
The complete picture at this point:
• Bot resolves the interaction → ZVA Survey widget fires the Bot Interaction Survey → interaction_type = bot_only
• Bot escalates to agent → Agent handles it → Queue-linked Agent Interaction Survey fires → interaction_type = bot_then_agent
• Customer goes directly to agent (no bot) → Queue-linked Agent Interaction Survey fires → interaction_type = agent_only
Every survey response now has both the right question language AND a machine-readable variable telling you exactly what path the customer took.
PHASE 6
Build Segmented Survey Reports in CX Analytics
Create reports that separate bot feedback from agent feedback using the Survey dataset and your custom interaction_type variable.
Step 6A: Create a Survey Report
1) Navigate to CX Analytics
Zoom Web Portal → Analytics & Reports → CX Analytics → + New report
Select the Survey dataset.
2) Add Widgets and Filters
Add report widgets for the metrics you care about:
• Summary widget: Response rate, completion rate (survey health check)
• Chart widget: Average rating by queue (compare bot vs. agent queues)
• Table widget: Individual responses (drill into specific feedback)
Click Add filter and filter by:
• Queue → [Your agent queue(s)] (Agent-handled survey results only)
• interaction_type → bot_only (Bot-only survey results only)
• interaction_type → bot_then_agent (Handoff survey results only)
• Channel → Voice / Messaging / Video (Channel-specific results)
3) Save and Schedule
Name the report (e.g., Survey Results by Interaction Type) and save it.
Consider creating three saved views: one filtered to bot_only, one to bot_then_agent, and one to agent_only — so stakeholders can quickly access the view relevant to them.
Step 6B: Use CX Insights for Exploratory Analysis
1) Ask Natural-Language Questions
Analytics & Reports → CX Insights
CX Insights supports natural-language queries against your contact center data. Try prompts like:
• “What are the top drivers of negative sentiment for bot-only interactions this month?”
• “Compare average CSAT for bot_only vs. agent_only interactions in the last 30 days.”
• “Which intents have the highest escalation rate from bot to agent?”
For best results, specify the time period, queue or team, and the breakdown you want.
Step 6C: View Individual Survey Responses
1) Access Response-Level Data
Admin Center → Content management → Surveys → Consumer engagement survey → [Select survey] → Results icon
Use Results by Respondent to see individual answers, or Results by Launch to filter by queue.
Since you created separate surveys for bot and agent interactions, each survey’s results page naturally contains only the relevant feedback type.
FINAL STEP
Validate End-to-End Before Going Live
Test every interaction path to confirm surveys fire correctly, variables are set properly, and reports show the right data.
Before trusting any survey data in production, run these validation scenarios:
| Test Scenario | Expected Variable | Expected Survey | Verify In |
| Bot resolves without escalation | bot_only | Bot Interaction Survey fires | Variables log + Survey results |
| Bot escalates to live agent | bot_then_agent | Agent Interaction Survey fires | Variables log + Survey results |
| Customer routed directly to agent | agent_only | Agent Interaction Survey fires | Variables log + Survey results |
| Multiple queue transfers | Value from initial path | Final queue’s survey fires | Variables log + Survey results |
Where to check:
✓ Variables Log
Contact Center Management → Logs → Variables
Set the time range, search by engagement ID, and confirm the interaction_type variable shows the correct value.
✓ Survey Results
Content management → Surveys → [Select survey] → Results
Confirm the test response appears under the correct survey (bot or agent).
✓ CX Analytics
Analytics & Reports → CX Analytics → [Your survey report]
Filter by interaction_type and confirm the test response appears with the correct segmentation.
Quick Reference: Complete Navigation Paths
| Task | Navigation Path |
| Create a survey | Admin Center → Content management → Surveys → Consumer engagement survey → + Create |
| Create custom variables | Admin Center → CX Management → Contact Center → Preferences → Variables |
| Enable variable for analytics | Preferences → Variables → [variable] → Edit → Use in CX Analytics and CX Insights |
| Edit a flow | Admin Center → CX Management → Contact Center → Flows → [flow] → Edit |
| Edit a ZVA bot flow | AI Studio → Virtual Agents → [chatbot] → Bot Flows → Edit |
| Link survey to a queue | CX Management → Contact Center → Queues → [queue] → Survey → Add |
| View Variables log | Contact Center Management → Logs → Variables |
| View survey results | Content management → Surveys → [survey] → Results icon |
| Build a CX Analytics report | Analytics & Reports → CX Analytics → + New report → Survey dataset |
| Use CX Insights | Analytics & Reports → CX Insights |
Once all six phases are configured and validated, you’ll have a survey infrastructure that answers the question every contact center leader should be asking: “What do our customers think about our bot versus what do they think about our people?” — with data clean enough to actually act on.
"This guide is based on Zoom Contact Center's current configuration as of November 2026. Navigation paths, feature names, and available settings may change with platform updates. Always verify against Zoom's official support documentation at support.zoom.com." This guide is based on Zoom Contact Center's current configuration as of November 2026. Navigation paths, feature names, and available settings may change with platform updates. Always verify against Zoom's official support documentation at support.zoom.com.
For additional technical insights tailored to IT admins and decision makers, visit the Zoom Technical Library.
Written by Roger Lee, Product Adoption Expert, Zoom Contact Center. For questions about this article or other Zoom Products, reach out via the Zoom Community or your Customer Success team.
