This section describes in detail the steps required to migrate from Legacy Digital Recommendations to Product recommendations feature of Acoustic Personalization.
If you are a customer who is not migrating from Legacy Digital Recommendations, and wishes to implement the Product recommendations feature from grounds-up, you should refer Product recommendations.
Legacy Digital Recommendations
For reference, here's the User Guide for Legacy Digital Recommendations.
Benefits of Acoustic Personalization
Salient features of Personalization as a product:
- Personalization improves the user experience of your team and your end-users through modern infrastructure, design, and Artificial Intelligence
- Business rules are applied to model results in real-time, an improvement from the daily updates you have today
- New and improved user interface that integrates content personalization, testing, and product recommendations capabilities into one personalization solution
- Cloud-powered and scalable to meet your needs
- Ongoing investment in new product features and machine learning models
- Flexible and powerful catalog ingestion service
Legacy Digital Recommendations comparison with Product recommendations
Legacy Digital Recommendations |
Product recommendations |
|
Product Recommendations |
✔️ |
✔️ |
A/B Testing for Prod Recs |
✔️ |
In Development |
Content Decisioning |
|
✔️ |
A/B Testing for Content |
|
✔️ |
Audience Import from Acoustic and Third-Party Products |
|
✔️ |
Acoustic Personalization uses Acoustic Exchange to give you more flexibility to use data from additional sources for personalizing product recommendations.
Enhancements in product recommendation models
Product |
Legacy Digital Recommendations |
Product recommendations |
View to View |
✔️ |
✔️ |
Buy to Buy |
✔️ |
✔️ |
View to Buy |
✔️ |
✔️ |
Abandon to Buy |
✔️ |
✔️ |
Most Popular |
✔️ |
✔️ |
Ability to Combine Models |
✔️ |
See Note 1 |
Model Targets or Inputs |
✔️ |
✔️ |
Business Rules |
✔️ |
✔️ |
Fallback Options |
Up to 6 backups |
Primary and Secondary |
NOTE 1: User feedback shows that merging multiple models is confusing, and they would prefer using one model that was more personalized to deliver relevant results, at a time.
Modernized model techniques
Models |
Legacy Digital Recommendations |
Product recommendations |
View to Buy |
Frequency-based |
Pattern-Mining, |
Abandon to Buy |
Frequency-based |
Pattern-Mining, |
View to View |
Frequency-based |
Collaborative filtering |
Buy to Buy |
Frequency-based |
Collaborative filtering |
Most Popular |
Frequency-based |
Analytics-based |
Personalization uses algorithms commonly used in the real-world by thought leaders in artificial intelligence and machine learning. Personalization models look for patterns between behavioral and catalog data, whereas the legacy product only takes into consideration the frequency of views and buys.
Migration workflow
Before you begin with the migration process, ensure that the prerequisites are fulfilled.
Migration is planned to be conducted in three phases.
Phase 1 consists of the following high-level steps:
- Onboard the customer organizations and users in Personalization
- Import Legacy Digital Recommendations product catalogs
- Collect behavior events
Phase 2 (optional) consists of the following high-level steps:
- Map Legacy Digital Recommendations Offers to Product recommendations strategies
- Map Legacy Digital Recommendations business rules to Product recommendations strategies business rules
- Serve Legacy Digital Recommendations using Product recommendations
Phase 3 consists of the following high-level steps:
- Integrate Acoustic Personalization Library
- Configure your site
- Test and launch
- Benefit from Personalization’s new features and enhancements
Migration flow is carefully designed to minimize code changes on the website during the Phases 1 and 2.