We're rolling out a truly transformative update: the launch of our AI-powered Audience/Segment Builder. This release also brings a key new Machine Learning model to our "On Your Data" (OYD) BigQuery users and significant stability improvements across the board.

🔥 What’s new

• [Beta] Build Audiences with AI using Natural Language Prompts!

  • Segmenting your customers is now as easy as asking a question. We've introduced the Audience/Segment Builder with Large Language Model (LLM) integration! Simply describe the audience you want in plain English (e.g., "Customers who bought a red item in the last 30 days but haven't engaged with an email"), and Wondaris will instantly generate the corresponding filters for you.

• New Predictive Model: Propensity to Churn (OYD - BigQuery)

  • We've expanded our Machine Learning capabilities for "On Your Data" (OYD) BigQuery users with the introduction of the Propensity to Churn model. You can now proactively identify customers at risk of leaving, allowing you to launch targeted retention campaigns and boost customer lifetime value. [Placeholder for a screenshot/video showing the new Churn model setup]

• Improved Clarity for ML Model Schedules

  • When viewing your Churn models, the Training schedule and Enrichment schedule are now clearly displayed in the model summary. This enhancement ensures you always have complete visibility into when your models are running and applying predictions.

🔧 Fixes and Improvements

Improvements

• Enhanced AI-Generated Filter Readability: We've improved the underlying logic of our LLM Service to include field aliases in the response, meaning the generated filters are now easier to read and understand in the Audience/Segment Builder.

• Simplified Multi-Value Selection: We've made small but effective UI tweaks to the selection box for multiple categorical values in filters, making it easier to select numerous options in one go, particularly in the Consent Rules settings.

Fixes

• Robust ML Model Training: We've resolved several issues where Propensity to Buy and pLTV models for Snowflake and BigQuery "On Your Data" (OYD) were getting stuck in the 'Training and Evaluating' status or failing unexpectedly.

• Data Modelling and Source Stability:

  • Resolved multiple 'Fatal Errors' that occurred when adding/editing tables or creating enrichment models, ensuring stable core data platform functionality.

  • Fixed issues where the Audience/Segment Count Daily Job wasn't triggering automatically in OYD, ensuring your data is kept up-to-date.

• User Interface and Workflow:

  • Fixed UI bugs related to status formatting, misaligned select boxes in Activation, and incorrect display of Custom Fields when a Consent Field was selected in GA4 mapping.

  • Corrected an issue where the Audience state was incorrectly marked as "dirty" after reverting filter changes, preventing unnecessary saving prompts.

  • Addressed numerous display and link issues with the new Wondaris Help Entity pop-ups to ensure documentation is correctly displayed and accessible across the Audience/Segment builder.

🤔 Good to know

Please contact your implementation partner or support@wondaris.com to discuss any queries you may have regarding enabling any new features within your Wondaris instance, especially the new [Beta] LLM Audience Builder.

We encourage you to explore these powerful new features and improvements. Your feedback is invaluable in helping us shape the future of Wondaris. Please don't hesitate to reach out with any comments or feature requests – we'd love to hear from you!