AI KPIs, Pivots, Measures

🤖 AI KPIs

📄 Overview

GermainUX provides AI KPIs that capture the results and performance of AI analysis across user sessions, errors, conversations, recommendations, and AI requests.

These KPIs make AI-generated insights available throughout GermainUX for dashboards, analysis, reporting, alerting, and automation.

📊 KPI Catalog

AI Error Analysis

AI Error Analysis contains an AI-generated analysis of detected error events.

Instead of reviewing individual error occurrences manually, AI can analyze the available error information and summarize what happened.

💡 Use this KPI to:

A typical investigation can progress from:

Error → AI Error Analysis → Affected Population → Related KPIs → Instances → Likely Root Cause

🎥 AI Session Analysis

AI Session Analysis contains an AI-generated analysis of a recorded real user session.

It summarizes significant activity within the session so teams can understand what happened without manually replaying the entire session.

🔍 Depending on the monitored data, the analysis can identify:

When additional context is required, teams can open the associated Session Replay to see exactly what the user experienced.

User Session → AI Session Analysis → Significant Events → Session Replay

❗ AI Session Friction

AI Session Friction contains friction identified by AI while analyzing a real user session.

Detected friction can include conditions affecting:

Examples can include confusing interactions, repeated actions, unsuccessful attempts, abandonment, errors, application slowness, or inefficient workflows.

These friction KPIs make it possible to aggregate findings across sessions and determine which problems affect the largest populations.

User Sessions → AI Session Frictions → Recurring Frictions → Affected Population → Analysis

✨ AI Session Recommendation

AI Session Recommendation contains recommendations generated by AI from a collection of analyzed user sessions.

Rather than focusing only on what happened in one session, this KPI can surface recommendations based on patterns observed across multiple sessions.

🔧 Recommendations can identify opportunities to improve:

For example:

Thousands of User Sessions → AI Analysis → Recurring Frictions → AI Session Recommendations → Prioritized Improvements

This helps teams move from identifying individual issues to determining what should be improved across a broader user population.

💬 AI Chat Friction

AI Chat Friction contains friction detected by AI within a monitored chat or conversation.

AI can analyze conversations involving customers, employees, agents, or bots to identify interactions where the conversation may not be progressing successfully.

Examples can include:

When chat data is associated with a monitored user session, teams can investigate the surrounding user experience for additional context.

Conversation → AI Chat Friction → User Session → Session Replay → Related Events → Analysis

⏱️ AI Guidance

AI Guidance measures the time required for GermainUX AI to process a question and return a response.

This KPI provides visibility into the responsiveness of AI Guidance itself.

📋 Use it to monitor:

This KPI measures AI service performance, rather than an AI-generated business or user-experience insight.

⚙️ GermainUX AI Request

GermainUX AI Request measures the duration of queries executed against GermainUX by AI.

It provides visibility into the time required to retrieve and process GermainUX data used by AI capabilities.

📈 Use this KPI to:

This KPI is primarily a technology-performance KPI supporting the monitoring and troubleshooting of GermainUX AI operations.

🌐 Analyze AI Insights at Scale

Because AI findings are stored as KPIs, GermainUX can analyze them individually or aggregate them across larger populations.

For example:

1,000 User Sessions
→ AI Session Analysis
→ AI Session Friction
→ Identify Recurring Frictions
→ Determine Affected Population
→ AI Session Recommendations
→ Prioritize Improvements

This allows teams to move from understanding one user's experience to identifying the most important issues and opportunities across thousands of users.

📁 From AI Insight to Evidence

AI KPIs can be correlated with the underlying data collected by GermainUX, including user sessions, application activity, errors, transactions, performance, and other KPIs.

A typical investigation can progress from:

AI Insight → Friction / Issue → Affected Population → Related KPIs → Individual Instances → Session Replay / Trace → Likely Contributing Factors

This allows teams to use AI to quickly identify what deserves attention while retaining access to the underlying evidence for further investigation.

🔁 Pivots

AI KPIs can be analyzed using pivots to understand where AI-generated insights, frictions, recommendations, or performance issues occur.

Available pivots vary by KPI and can include:

For example:

AI Session Friction → Application → User → Session → Session Replay

can help identify which applications generate the most friction, which users are affected, and the individual sessions where the friction occurred.

Pivots by AI KPI

The exact pivots available depend on the AI KPI and monitored data.

🔢 Measures

Measures provide the numerical values available for analyzing AI KPIs.

KPIs without a dedicated numerical measure can still be aggregated, counted, filtered, and analyzed using their available pivots.

🔎 Example Analyses

Combining AI KPIs with pivots and measures makes it possible to answer questions such as:

These combinations allow teams to move beyond an individual AI result and determine what is happening, where it is happening, who is affected, how frequently it occurs, and how it changes over time.

📦 Additional Resources

• AI Management

• AI Advice Dashboard

• AI Summary for Real User Session

• Frictions Dashboard

• More Analytics Features

• More Automation Features

• Dashboard & Reports

• Management Features.

ℹ️ Get Help

The Germain Team can help you set this up. Contact GermainUX Support.