Smart Analyst

Use Smart Analyst when you need to understand why calls are failing, escalating, or taking too long — without pulling data manually or writing queries. It surfaces patterns from recent conversations so you can prioritize what to fix next. Smart Analyst uses AI to review recent conversations and answer natural-language questions like “What are the top escalation reasons?” The analyst appears as a chatbot inside Studio, analyzing transcripts and metadata to surface insights on agent performance, customer sentiment, and conversation patterns. You can query data from the previous day or specify a custom date range.

​

What you can do

• Ask freeform questions about conversation data
• Identify top containment blockers and transfer reasons
• Surface technical issues, drop-offs, or sentiment trends
• Prioritize what to fix next without pulling data manually

​

Deep sampling

Smart Analyst analyzes up to 500 conversations per query. Conversations are sampled randomly or based on a custom metric / PolyScore, giving you the flexibility to explore broad trends or investigate specific behaviors at scale.

​

What you see during deep sampling

When Smart Analyst runs a deep sampling query, you see the analysis unfold in real time:

1. Planning — Smart Analyst shows which steps it will take to answer your question.
2. Reasoning steps — Live updates appear as the analyst searches data, queries metrics, or retrieves conversation details.
3. Sampling progress — A progress bar tracks how much of the sample has been analyzed (e.g. 40%, 70%, 100%).
4. Final answer — Once the full sample has been analyzed, Smart Analyst delivers a structured response with its findings.

This staged approach gives you visibility into what the analyst is doing and how far along it is — you don’t have to wait for a response with no feedback.

​

Sampling modes

Smart Analyst supports two sampling modes. The mode is chosen automatically based on your query and the available data.

​

Improving deep sampling results

The quality of your deep sampling results depends on how well your project is configured and how specific your prompts are. Define clear custom metrics. Custom metrics are used both for filtering and for enriching each conversation’s metadata during analysis. The more precise your metric definitions, the better Smart Analyst can target relevant conversations and interpret their outcomes. Vague or overlapping metrics lead to noisier samples. Be specific in your query. Rather than “What’s going wrong?”, try “What are the top 5 reasons calls are handed off to a human agent in the last 30 days?” — this gives the analyst a clear objective and time frame. Combine metrics with natural-language context. If your project has a metric like HANDOFF_REASON, you can ask: “Among calls where HANDOFF_REASON is ‘billing’, what patterns do you see in the conversation flow?” This combines metric-based filtering with transcript-level analysis.

​

Understanding sample coverage

Deep sampling analyzes a maximum of 500 conversations per query, covering a date range of up to 90 days (defaulting to the last 30 days). For most projects, 500 conversations should be enough to provide strong directional insight — enough to identify recurring patterns, common failure modes, and behavior frequency.

​

Starting from a dashboard

You can launch Smart Analyst directly from the Home dashboard. The Generate insights and Create analysis buttons on each chart open Smart Analyst with a prompt already populated based on that chart — for example, clicking Create analysis on the containment rate chart opens a query scoped to containment patterns. This is the fastest way to investigate a trend you’ve spotted in the dashboard without having to write a query from scratch.

​

Tips for better results

Smart Analyst produces the most relevant insights when you include project context in your prompts — such as your domain, goals, and key terminology.

​

Provide project context

Instead of:

“What are the top transfer reasons?”

Try:

“This is a hotel reservation agent. The main goal is containment rate. What are the top reasons calls are transferred to a human agent?”

Adding context about your domain, goals, and terminology helps Smart Analyst interpret the data more accurately.

​

Ask for percentages explicitly

For precise quantitative breakdowns, include percentage or count-based phrasing in your prompt:

“Please give me the percentage breakdown of transfer reasons.”

This produces more structured, data-driven answers compared to open-ended narrative questions.

​

Use it as a call review starting point

A common workflow is to use Smart Analyst to identify problem areas, then drill into individual conversations:

1. Ask Smart Analyst: “Show me five calls from the last week that couldn’t be answered.”
2. Review the flagged conversations in Conversation Review.
3. Use the findings to update your knowledge base or adjust flow logic.

This is especially effective as a daily or weekly optimization routine.

​

Example prompts

Smart Analyst understands natural questions across a range of categories. Here are examples across different verticals:

​

Containment and transfers

• What are the top 5 reasons conversations are transferred to a human agent?
• What percentage of calls are contained vs. handed off?
• What actions could increase containment for lookup failures?

​

Quality and sentiment

• Do customers express frustration with the agent?
• What types of conversations result in negative sentiment?
• Are customers complaining about wait times or hold durations?

​

Conversation analytics

• Which conversations had the longest average handling time yesterday?
• Which intents failed the most often?
• What are the most common caller requests this week?

​

Knowledge gaps

• What kinds of questions are we not handling well?
• What information does the agent usually request from callers?
• Where does the agent give incorrect or incomplete answers?

​

Domain-specific examples

• What are the top reasons delivery status lookups fail? (logistics)
• How often do callers ask about appointment rescheduling? (healthcare)
• What are the most common billing questions? (utilities / financial services)
• What drives drop-offs in the reservation flow? (hospitality)

​

Templates

Pre-built templates help you start with ready-made prompts grouped by Customer Insights, Agent Performance, and Custom Research.

• Each template includes a description and an optimized query
• Selecting a template populates the prompt field — run it or modify as needed
• Templates are especially useful if you are not sure where to start, or if you are monitoring performance on your own and want a structured starting point

​

How Smart Analyst works

​

Related pages