Chatpulse® – Go beyond Copilot Studio analytics.

Understand what is happening across your Microsoft Copilot virtual assistants and AI agents – and why.

Chatpulse analyses every single Copilot conversation to uncover performance issues, quantify their impact and show where to improve.

Chatpulse - chatbot analytics workbench screen

Uncover what standard Copilot metrics miss.

  • Reduce avoidable agent handovers and operational cost

  • Identify failed journeys, fallbacks, misrouting and abandonment

  • Improve customer experience and conversation outcomes

  • Prioritise fixes by scale, impact and business value

  • Measure the effect of changes over time

Chatpulse - chatbot analytics workbench indicator metrics

Get to the heart of every Copilot conversation with Chatpulse.

Copilot Studio metrics typically metrics show what happened, but rarely explain why performance changed or where to act.

Chatpulse analyses complete conversation data from Copilot agent conversations to uncover root causes, quantify their impact and prioritise the changes with the greatest value. Crucially, it looks for conversational markers that make a bad experience.

See the patterns.
Understand the conversations.

Chatpulse combines quantitative performance analytics with qualitative analysis of what customers actually said.

Explore trends, topics, journeys, cohorts and individual conversations in one place. Move from an executive view of your bot’s performance to the exact messages behind an issue.

Use the evidence to improve Copilot agent performance and measure whether each change worked.

Chatpulse - chatbot analytics example metrics

  • Find the topics and journeys driving fallbacks and failed outcomes

  • Identify customer intents that are being misrouted or misunderstood

  • Understand where and why users abandon a chatbot or voicebot journey

  • Investigate avoidable transfers and escalations to human agents

  • Group related utterances and quantify issues by volume and impact

  • Drill down from Copilot performance trends to individual messages

  • Compare cohorts, channels and time periods to isolate changes

  • Measure improvements and demonstrate their operational and customer value

Conversational AI analytics for Copilot Studio that turns data into action

Chatpulse is a powerful conversational analytics platform for uncovering, troubleshooting and sizing issues across automated conversations. It works seamlessly with virtual agents built in Microsoft Copilot Studio.

Explain why performance changed

Connect changes in chatbot performance to the topics, journeys and messages behind them.

Analyse every conversation

Review complete datasets at scale using Chatpulse’s proprietary analytics, classifiers and language models.

Find hidden failure patterns

Group semantically related utterances to reveal recurring issues that normal fallback counts and NLU scores can obscure.

Prioritise by business impact

Quantify the volume, customer effect and operational cost of each issue so teams know what to fix first.

Compare cohorts and periods

Compare channels, regions, bot versions, customer groups and time periods to isolate what changed.

Investigate complete journeys

Move from portfolio-level trends to individual conversations and messages without losing the surrounding context.

Connect Copilot Studio data

Import conversation data from Microsoft Copilot Studio and Dataverse and make sense of it.

Track improvement over time

Validate whether releases, content changes and routing fixes produced the intended customer and operational result.

Support shared decisions

Give product, conversation design, CX, operations and leadership teams evidence they can act on together.

Chatbot analytics in action

Misrouting identified at scale

By grouping shipping conversations and comparing where they were routed, Chatpulse found that 31% were reaching pre-sales agents. Correcting the journey reduced hundreds of avoidable agent calls each month.

Demand hidden inside sentiment

A 10% rise in negative sentiment on New Year’s Day exposed demand that neither the chatbot nor the unavailable agent team could resolve. The evidence gave the business a clear plan for the next peak.

A translation issue affecting thousands

Conversation-level analysis revealed that a translated voicebot greeting was confusing thousands of callers. Analysts could isolate the wording, understand the pattern and rephrase the message.

A handover loop revealed

Chatpulse identified hundreds of conversations repeatedly offering an agent without completing the transfer. Quantifying the loop made the problem visible and the underlying bug was fixed.

Chatpulse logo

Taking Microsoft Copilot Studio analytics to the next level

Copilot Studio provides operational reporting. But Chatpulse adds the deeper conversational context needed to understand the topics, journeys, failure points behind those results.

Through the Chatpulse connection with Dataverse, teams can investigate complete conversations, compare cohorts and time periods, quantify misrouting, escalation and abandonment, and measure whether each improvement delivered the intended outcome. Chatpulse can also accept additional data augmented to the conversation transcripts, to enable more advanced filtering and provide a wider business context for analysis.

Use Chatpulse your way.

Chatpulse pricing is tailored to conversation volume and the support you need. Use the platform independently or combine it with expert-led analysis, regular reporting, bespoke investigations and recommendations through Chatpulse Conversational Analytics as a Service.

Chatpulse - chatbot analytics workbench indicator metrics

Chatbot analytics frequently asked questions

Answers to common questions about Chatpulse, conversational AI analytics and improving chatbot performance.

Chatbot analytics is the process of measuring and analysing automated conversations to understand performance, customer behaviour and where journeys can be improved. Chatpulse combines quantitative measures with analysis of the language used in complete conversations.

Useful measures include goal completion, containment, escalation, abandonment, fallbacks, customer sentiment and operational cost. Chatpulse goes further by identifying the topics, journeys and messages that explain why those measures changed.

Copilot Studio reporting provides an operational view of agent activity. Chatpulse adds deeper conversation-level analysis so teams can investigate the causes behind performance, quantify recurring issues and prioritise improvements.

Yes. Chatpulse can connect to Microsoft Copilot Studio conversation data through Dataverse, allowing teams to analyse complete conversations, associated metadata and performance trends.

Chatpulse applies proprietary analytics, classifiers and language models to Copilot Studio virtual agent transcript datasets. It maintains information on key conversational indicators, allowing analysts to effectively segment the conversations and discover issues and pain points, as well as monitor improvements.

Chatpulse is platform agnostic. It can analyse well-structured conversational data supplied in formats such as CSV or JSON and supports direct connections for selected platforms, including Microsoft Copilot Studio through Dataverse.

Yes. Chatpulse can analyse text from chatbot and voicebot transcripts, virtual assistants and AI agents – provided the conversation data and relevant metadata are available.

Timing depends on the data source and the quality of its documentation. For well-structured conversational data, such as Microsoft Copilot Studio transcripts through Dataverse, onboarding and mapping are normally straightforward.

Data handling, access and retention requirements are agreed during onboarding. Personally identifiable information can be redacted before analysis, and further details are available in the Chatpulse Terms of Service.

Yes. Teams can use the platform directly or combine it with Chatpulse Conversational Analytics as a Service, including expert-led analysis, regular reporting, bespoke investigations, recommendations and impact tracking.

Talk to us about chatbot analytics

Tell us about your chatbot, Copilot Studio agent, voicebot or virtual assistant and what you want to improve.

  • info@thecaicompany.com

  • The CAI Company, Saltford House, High Street, Saltford, Bristol, BS31 3ED, UK.