Advanced analytics
Advanced conversation analytics
Turn a stream of conversations and inbound messages into measurable metrics and clear dashboards. Build charts from saved views and filters, or describe a series in plain language — the AI builder figures out which records to count. Search by meaning instead of exact words, slice data by operator, topic, tag and speaker, add derived time metrics and export everything to your database or through the REST API and MCP. Data refreshes roughly every 5 minutes — this is managed analytics across your whole archive, not real-time BI.
Contact topics, this week
ContactsKey facts
- Dashboards
- 10 chart types (line, area, bar, stacked, horizontal bar, scatter, radar, pie, doughnut, polar); period comparison, normalization, grouping by day/week/month.
- Series from views
- Each chart series is built from a saved view or an inline filter — you count exactly the slice of conversations you need.
- AI series builder
- Describe in words which records to count, and the AI assembles the filters, metric and aggregation for you.
- Semantic search
- Search by meaning across the whole archive — it surfaces relevant conversations and messages even when the exact query words are absent.
- Export and databases
- Auto-sync to PostgreSQL, MySQL, MongoDB and ClickHouse, plus a REST API and MCP server to work from code and AI agents.
Dashboards for your questions
Build charts that answer specific business questions — from the same data that's already scored with metrics and topics.
- 10 chart types: line, area, bar, stacked, horizontal bar, scatter, radar, pie, doughnut and polar.
- Period comparison, normalization and grouping by day, week or month in one click.
- Every series is built from a saved view or an inline filter — you count exactly the slice you need.
A chart series in words
No need to wire up filters and aggregations by hand: describe the task in text and the AI builder assembles the series itself.
- "Show the share of negative contacts by week" — the AI picks the metric, filters and chart type.
- The result is transparent: you see which records went into the series and how they were aggregated.
- It's easy to fine-tune by hand — the AI gives you a starting point, not a black box.
Search by meaning and slices
Find the conversations you need by meaning and cut the data along any dimension that matters to your team.
- Semantic search: relevant conversations and messages are found even when they don't contain the exact words.
- Slices by operator, topic, tag and speaker — compare people, lines of business and periods.
- Derived time metrics (hour, day of week, month) turn conversation time into an analytical dimension.
Export, databases and API
Analytics isn't locked in — push data to where your team works and build reports in your own tools.
- Auto-sync of metrics and records to PostgreSQL, MySQL/MariaDB, MongoDB and ClickHouse.
- A REST API (v1) and MCP server — pull data from your own code and from AI agents.
- Data refreshes roughly every 5 minutes — an honest cadence, not a real-time promise.
Product
Advanced conversation analytics
- 1
Score the conversations
Conversations and messages are scored automatically with system and custom metrics, topics and tags — the foundation for analysis.
- 2
Build a dashboard
Add series from saved views, or describe them in words — the AI builder assembles the filters and aggregation.
- 3
Slice and compare
View slices by operator, topic, tag and speaker, compare periods and find the conversations you need by meaning.
- 4
Export it out
Connect an auto-sync to your database or the REST API and MCP to build reports and integrations in your own tools.
FAQ
Frequently asked questions
How fresh is the data in dashboards?
Analytics data refreshes roughly every 5 minutes — a regular sync, not real-time BI. That cadence is plenty for quality control, reporting and trend analysis, and we honestly don't promise sub-second latency.
What does "a chart series in words" mean?
Every series on a dashboard has an AI builder: you describe in text which records to count and which metric to show, and it assembles the filters, aggregation and chart type. The result is transparent and editable by hand — it's a starting point, not a black box.
How is semantic search different from keyword search?
Semantic search finds relevant conversations and messages by the meaning of your query, even when the exact words are missing: "complaints about turnaround" also surfaces "waited a long time" and "didn't make it in time." It's available both in filters and in the copilot.
Which dimensions can I slice data by?
By operator, topic, tag and speaker, as well as by derived time metrics — hour, day of week, month. This lets you compare people, lines of business and periods and pinpoint exactly where a metric is dropping or rising.
Can I export analytics into my own system?
Yes. There's an auto-sync of metrics and records to PostgreSQL, MySQL/MariaDB, MongoDB and ClickHouse, plus a REST API and MCP server — you pull the data into your own code, BI tools and AI agents and build reports wherever it suits you.
Related capabilities
Related use cases
Ready to start?
Turn every conversation into data, knowledge and action
Start by analyzing your conversations — no risk, no bots required. The platform turns your archive into data and a knowledge base, and voice agents plug in when you're ready.
Free plan, no card required.