Concept: AI Topics

Workday Peakon's AI Topics uses Natural Language Processing (NLP) to analyze employee feedback and identify company-specific topics by clustering comments based on meaning rather than using pre-defined categories. It then uses AI to summarize these clusters (groups of comments), highlighting sentiment, volume, key segments, and associated theme, providing context. This provides organizations with granular, actionable insights directly within Peakon, helping leaders to identify trends and address key challenges and opportunities.

Topics are automatically calculated at the close of each survey round. The calculation groups comments into themes for each segment based on the aggregated data at that time.

Administrators can manually recalculate topics when needed. For more information, visit Manual Topic Regeneration. Topics are calculated for manager and company contexts only. If you wish to generate topics automatically for other attributes, you can add them under

Administration

Data Settings

Topics

Attributes for Topics Generation

.

AI topics are:

Company-specific: The system uses Workday-developed Natural Language Processing (NLP) to analyze comments from each survey round. It groups comments into clusters based on their semantic meaning, rather than relying on predefined categories.
Insightful: Peakon uses Natural Language Processing to extract real meaning from comments. Example: Low priority on common words with little or no meaning in answers and questions.
Summarized: Uses Large Language Models (LLMs) to generate a title and a summary for each topic.
Contextualized: Each topic is enriched with additional data points, including sentiment, comment volume, the key segments impacted, and a broader, overarching theme. Highlighted based on actionability, enabling leaders to focus on what's most important to their teams. Available in all survey languages.
Available in all survey languages when Comment Translations is enabled.

To support high-level analysis, individual topics are grouped into broader, sentiment-agnostic themes that capture shared underlying meanings (Example:

Management, Workload

). In each survey round, Peakon highlights the top 3 themes based on the prominence of associated topics. This categorization enables you to explore patterns and recurring areas of focus across survey rounds.

Key segments show where a specific topic is most relevant. These are not necessarily the largest segments. Instead, they are the segments where a topic is statistically overrepresented compared to the rest of the segments present in the round.

This helps you identify segments of interest or concern that might otherwise be hidden in larger groups.

Topics are available on the

Insights

page for all managers with access to comments. Peakon only runs topic calculations when a survey round contains 10 quality comments and at least 50 comments present in the Peakon instance across all the rounds. Once this global threshold is met, additional thresholds determine the minimum topic size:

A segment qualifies for topics when it meets all of the following:

Each topic available in the

Topics

page includes: