Understanding users and their behaviors is essential to delivering exceptional support experiences.
As customer expectations rise, businesses need more precise, actionable insights to guide strategy, optimize performance, and continuously improve the quality of their interactions.
That’s why we’ve introduced a comprehensive set of upgrades to our Customer Intelligence experience—making it more powerful, more transparent, and more insightful than ever before.
In this blog post, we’ll walk through all the updates, explain how they strengthen your ability to interpret user conversations, and show how these tools support better decision-making across your support organization.
A Refined Focus for Customer Intelligence
The Customer Intelligence feature has always offered valuable insight into how users interact with your agent. With this update, we’re taking it several steps further.
You now have more detailed information about the people behind each query, the environment they’re using, and the context in which conversations occur.
These updates make it easier to diagnose issues, understand user needs, and build a clearer picture of customer behavior patterns.
New and Improved Filters for Deeper Insights
To make it easier to refine and analyze your data, we’ve added several new filter options. These filters allow you to narrow down conversations with precision, helping you focus on the exact interactions and behaviors that matter most.
New filters include:
- Anonymous vs. Team Member Status: Quickly understand whether the query came from an unidentified user or someone on your internal team.
- CRM ID: Filter by External ID for faster cross-referencing or user-specific investigations.
- Query Sent By: Focus on queries sent by team members to separate internal testing or troubleshooting from customer interactions.
- User Location: Filter conversations by the user’s country to identify geographic patterns or region-specific needs.
- Deployment: Analyze interactions by deployment type, allowing you to evaluate performance across different agent configurations.
All previously missing information has been added to the modal, and every new field is also available in data exports—ensuring that your reporting remains complete and consistent across tools.
These updates make it significantly easier to slice and segment conversations for research, performance reviews, quality checks, or strategic evaluation.
Advanced Analytics: Getting More from Every Interaction
We’ve significantly strengthened the analytics depth within Customer Intelligence. Our proprietary analytics agent now extracts advanced quantitative and qualitative metrics from every query, giving you insight into customer behavior and AI blind spots that goes well beyond keywords or sentiment.
These advanced metrics help teams identify opportunities, track emerging issues, and understand whether the agent is performing as expected.
Content Source Found
This metric identifies whether the agent relied on a relevant content source to generate its response. It helps you measure how effectively your knowledge base supports the agent and can signal when additional documentation might be needed.
User Emotion
The system now detects the emotional tone of user messages across categories such as:
- Positive
- Neutral
- Frustration
- Dissatisfaction
- Confusion
Understanding user emotion enables proactive improvements and better customer experience management.
User Intent
Each query is classified by purpose, helping you understand why the user reached out. Supported intent categories include:
- Informational
- Troubleshooting
- Instructional
- Greetings
- Transactional
- Navigational
- Follow-up
This clarity supports more targeted optimization of both workflows and documentation.
Language Detection
User queries can come in many languages. The analytics agent now identifies the language used so you can better tailor support resources and segment multilingual audiences.
Fidelity
Fidelity tracks whether the agent remains consistent with its assigned persona. Low fidelity or frequent deviations can indicate that the persona configuration needs refinement.
Jailbreak Detection
Attempts to circumvent system rules—successful or failed—are now automatically tracked. This ensures you have visibility into risky or suspicious behavior.
Prompt Leakage
This metric monitors attempts by users to extract system-level instructions or hidden prompts. Identifying patterns helps reinforce security and improve safety configurations.
Profanity Classification
Customer Intelligence also flags NSFW or abusive user messages blocked by filters, giving you a clearer understanding of how often users engage in inappropriate behavior.
To switch between these advanced metrics, simply select the preferred Focus option from the menu. Having these metrics readily accessible makes it easier to interpret trends and identify where additional improvements or safeguards may be needed.
For more detail, you can refer to the full Advanced Metrics guide, which provides deeper explanations and use-case examples.
Powerful Search Capabilities
The search bar allows you to look for specific keywords in both user queries and agent responses. By default, the search engine returns conversations that include any combination of the words you enter.
To search for a precise phrase, wrap it in quotation marks: “your search keywords” This allows for highly targeted research—whether you’re reviewing product feedback, identifying recurring issues, or pulling examples for training.
Final Thoughts: A More Complete, More Insightful Customer Intelligence Experience
The latest enhancements to Customer Intelligence mark an important step forward in how teams understand and optimize user interactions.
By combining richer user details, advanced behavioral metrics, and more precise filtering tools, the platform now offers a clearer, more comprehensive view of every conversation.
These improvements empower support teams, product owners, and decision-makers to identify trends earlier, diagnose issues faster, and make smarter, data-driven adjustments that elevate the customer experience.
With deeper insights and greater transparency, Customer Intelligence continues to evolve into a powerful foundation for continuous improvement—helping you build more intuitive, responsive, and effective support processes at every touchpoint.
Build a Custom GPT for your business, in minutes.
Drive revenue, save time, and delight customers with powerful, custom AI agents.
Trusted by thousands of organizations worldwide
Frequently Asked Questions
How can I tell why the agent gave a wrong or incomplete answer in Customer Intelligence?
Chicago Public Schools reached a 91% AI success rate across 13,495 HR queries by finding failure points quickly. In Customer Intelligence, start with Content Source Found to see whether the agent retrieved relevant material. If no source was found, the issue is usually missing, outdated, or poorly connected knowledge. Then use filters such as Deployment, CRM ID, Query Sent By, Team Member Status, and User Location to see whether the problem is tied to a specific setup, tester, user, or region.
Which Customer Intelligence filters should I use first to find real support trends?
GPT Legal has handled 19,000+ legal queries from 5,000+ monthly visitors, which makes unfiltered analytics hard to act on. A practical starting order is to exclude internal traffic with Query Sent By and Team Member Status, then compare conversations by Deployment, then narrow to a specific person or account with CRM ID, and finally use User Location to spot regional patterns. That sequence separates testing noise from real customer behavior before you look for broader trends.
What should I check first in Customer Intelligence right after deploying an agent?
Integrity Data Insights LLC said, u0022Within a few minutes, they had a working chat bot.u0022 Right after launch, start by separating internal tests from real customer conversations using Query Sent By and Team Member Status. Then review Deployment and CRM ID if you need to isolate a specific configuration or user. Early datasets can be thin, so limited insight usually means you need more live conversations before trend analysis becomes useful.
Why might some interactions seem missing in Customer Intelligence?
TaxWorld has processed 189,351 queries at a 97.5% success rate, but analytics views still depend on how conversations are filtered. If an interaction seems missing, first review the filters you applied, especially Deployment, Query Sent By, Team Member Status, CRM ID, and User Location. Also compare what you see in the conversation modal and exports, since the update added previously missing information to both views. In many cases, the interaction is filtered out rather than absent.
Does the way I add knowledge affect conversation analysis?
VdW Bayern DigiSol trained WohWi AI on 3,620 documents and 25 million tokens, then saw a 50 to 60% task time reduction across 500+ member organizations. In Customer Intelligence, Content Source Found helps you see whether the agent is retrieving relevant material. That means source quality and freshness affect what shows up in analysis: incomplete or outdated knowledge can lead to weaker answers and different conversation patterns. Use analytics to spot repeated no-source conversations and then review the underlying content collection.
How can I verify incorrect emotion, intent, or other tags in Customer Intelligence?
CustomGPT.ai outperformed OpenAI in a RAG accuracy benchmark, but tag-level analytics still need review when results look off. Start by isolating a pattern with filters such as CRM ID, Query Sent By, User Location, and Deployment. Then use the added modal fields and exports to capture representative examples for internal review or support follow-up. Looking at repeated examples is more useful than judging a tag from one conversation.
Related Resources
These articles expand on the tools, workflows, and trust considerations behind stronger customer intelligence with CustomGPT.ai.
- Judgment-Free AI Bots — Explore how unbiased AI conversations can surface more honest customer insights and improve intelligence gathering.
- Context-Aware Agents — Learn how context-aware agents help CustomGPT.ai deliver more relevant responses based on prior interactions and business data.
- Priority Queries Live — See how priority queries help teams surface urgent requests faster and act on high-value customer signals in real time.
- Trust and AI — This piece examines the role of trust in AI adoption and why it matters when building reliable customer intelligence systems.