Skip to main content
The Audience Insights dashboard gives you a structured view of who your users are, what they are talking about, and how well your AI is categorizing their intent. Rather than sifting through raw conversation logs, you get precomputed clusters, topic rankings, and qualification metrics that translate directly into targeting decisions and bot improvement actions. Use this page regularly to stay aligned with shifting user behavior and to prioritize your team’s response efforts.

​
User Clustering

Neuroncx AI automatically segments your user base into clusters so you can identify patterns across large volumes of interactions without manual tagging. Three cluster types are available, each built on a different segmentation logic.

CategoryDefinedCluster

Groups users based on a set of predefined audience traits — such as patient type, service interest, or account tier — that are configured during your platform setup. These clusters reflect the standard segments most relevant to your business model.

UserDefinedCluster

Built on custom segmentation logic that you define. If your organization has specific rules for grouping users — for example, by insurance provider, preferred communication channel, or lead score threshold — those rules drive this cluster type.

LocationCluster

Organizes users by geographic attributes such as city or clinic location. This is especially valuable for multi-location healthcare and service businesses that need to understand demand and engagement at a regional level.
If you notice a cluster is unexpectedly small or empty, verify that the underlying trait or rule is correctly mapped to your incoming user data. Misconfigured attribute names are the most common cause of sparse clusters.

​
How to act on cluster data

Once you understand which clusters are growing or shrinking, you can tailor outreach accordingly. For example, a growing LocationCluster for a specific clinic could prompt you to schedule more availability there, while a spike in a UserDefinedCluster for high-intent leads might trigger an immediate follow-up sequence.

​
Topical Groupings

Beyond demographic clustering, Neuroncx AI groups users by the themes and topics that appear most frequently in their conversations. This gives you a content-level view of what your audience cares about right now. Common topical groupings include:
  • dr_profiles — users whose queries focus on finding, comparing, or learning about specific doctors or practitioners
  • appointment_related — users actively engaged in scheduling workflows, including booking new appointments, modifying existing ones, or asking about availability
  • high_value leads — users whose interaction patterns suggest strong purchase or conversion intent, typically based on query depth and engagement frequency
Topical groupings are derived from your live conversation data, so they evolve as your user base grows. Check them weekly to catch emerging themes before they become support bottlenecks.

​
Using topical groupings for better targeting

Each topical group maps to a distinct communication strategy. Users in the appointment_related group respond well to proactive reminders and friction-reduction messages. Users flagged as high_value leads benefit from personalized follow-ups or escalation to a human agent. Knowing which group a user belongs to helps your team deliver the right message at the right moment instead of relying on one-size-fits-all broadcasts. The Trending Topics panel displays the top 10 topics surfaced from recent user queries and interactions. This ranked list updates based on your selected date range and gives you a real-time pulse on what your users are asking about most. Typical trending topics include:
A high volume of TopicAwaited entries indicates that a significant share of user queries is falling outside your bot’s trained intent categories. Prioritize reviewing these queries and adding them to your training set to reduce unanswered or misrouted interactions.
Use the trending topics list as a weekly agenda item for your bot training and content team. If a topic is trending upward but your bot’s response coverage for it is low, you can create or refine intents to address it directly. If a topic is trending due to an operational issue — like a surge in cancellations after a scheduling outage — you can coordinate a response across support and marketing teams.

​
Topic Qualification

Topic Qualification measures how effectively your AI is categorizing the queries it receives. The section presents a metric-driven bar chart that breaks down the volume of queries by qualification status — showing how many have been successfully assigned to a recognized intent category versus how many remain unqualified.
Each bar in the chart represents a topic or intent category. The height of the bar reflects the number of queries assigned to that category within your selected date range. A well-trained bot will show healthy volumes across multiple categories with a relatively small residual in the unqualified bucket. A bot that is under-trained will show the opposite — a large unqualified segment and sparse coverage across defined intents.

​
What good qualification looks like

A high qualification rate means your bot is accurately recognizing user intent, routing conversations correctly, and generating relevant responses. This directly improves CSAT scores and reduces the need for human escalation. If your qualification rate is below expectations, use the unqualified query samples to identify the most common gaps and update your bot training accordingly.
Topic Qualification is one of the most actionable metrics in the Audience Insights dashboard. Review it after any major bot update to confirm your changes had the intended effect on categorization accuracy.