Analytics

Engagement Layer / Analytics

Intro

In the AI-driven engagement layer, data visualization transitions from being a static reporting tool to a dynamic component that adapts to the user's context. The goal is to minimize cognitive load by providing the right amount of information at the right time. Design requirements should be mapped based on the time to insight, an AI-driven decision-making process that determines whether a user needs a high-level trend or a detailed analysis.

AI insights annotations

The Chart Framework: Scaling Complexity

The engagement layer utilizes three distinct levels of chart complexity to maintain a clean UI and optimize the so-called user's time to insight across areas like conversations, spaces, and agents.

L1 (Micro) – Scanning: L1 visualizations are compact "spark-style" charts that provide immediate visual context for a single metric or trend without axes or labels. They are ideal for instant recognition within high-density areas, such as a discovery feed or a small widget in a space. User interaction is not available at this level.
L2 (Mini) – Understanding: L2 charts introduce basic axes, labels, and legends to offer context and aid in comprehension of specific insights. Interaction is limited to basic hover states or inspection overlays to reveal data points. L2 is suitable for conversations or agent responses where users need to validate AI-generated claims without switching contexts.
L3 (Full) – Analysis: L3 offers a comprehensive analytical view with full interactivity, including filtering, drilling down, and zooming. This level is used when users require complex exploration and is the primary setting for advanced AI annotations and comparative data analysis.

<div> <div>Level</div> <div>Goal</div> <div>Interaction</div> <div>Best For</div> </div> <div> <div>L1</div> <div>Instant Trend</div> <div>None</div> <div>Discovery, high-density lists</div> </div> <div> <div>L2</div> <div>Contextual Insight</div> <div>Hover/basic overlay</div> <div>Conversations, agent summaries, AI chat responses, preview cards</div> </div> <div> <div>L3</div> <div>Deep exploration</div> <div>Full (drill-down, filter)</div> <div>Spaces, root cause analysis</div> </div>

AI Annotations: The "Why" Behind the Data

AI Annotations add a layer of intelligence to our standard L3, and occasionally L2, charts. They help explain "what" the data is and "why" it matters. These annotations are represented as insight dots or lollipops placed directly on specific data points, such as a spike in a line chart or an outlier in a bar chart.

Insight Dot and Popover: The insight dot acts as a visual anchor for AI-generated insights. When users interact with this dot, an AI insight popover appears, providing a concise natural language explanation of the data anomaly or trend. For example, "Sales are 42% above target due to the Q3 Enterprise campaign."
Mapping Requirements to Annotations: When designing features, consider whether a data point needs an explanation beyond just a value. If the system detects a variance, anomaly, or milestone, use an AI Annotation to highlight these elements:
- Highlight Anomalies: Draw attention to unexpected shifts in data.
- Provide Root Cause: Use the popover to reveal the AI's reasoning.

information

When implementing these principles, prioritize the time to insight. If an L1 sparkline meets the user's needs, avoid defaulting to an L3 chart. Use AI annotations strategically to minimize the need for manual data interpretation, ensuring that the engagement layer remains proactive, intelligent, and seamlessly integrated into the SAP ecosystem.