Smarter Scheduling Recommendations + Easier Product Updates

What changed

1) Scheduling recommendations are now stricter and clearer

When Prism suggests a dayparting schedule (when to lower bids or pause ads by hour), it now applies stronger checks before showing that suggestion.

• When an Amazon Seller account was first linked, the system imported the previous 60 days campaign data. This is summarized by day, therefore cannot be used to recommend dayparting scheduling.
• The system now imports the previous 2 weeks of hourly campaign data (the max available), combining recent real performance data with older modeled history for a fuller 60-day view.
• It still requires enough (20 days) of real observed coverage before showing a recommendation. Now that the past 14 days of hourly data are imported, only 6 more days need to be collected before recommendations are generated.
• It still requires enough meaningful click activity to avoid weak signals.
• If a recommendation is skipped, the reason is now explicit (for example, not enough observed coverage yet).

You will also see clearer context in the recommendation details, including:

• how much data was observed vs modeled,
• the observed coverage percentage,
• the minimum coverage threshold,
• whether that threshold was met.

2) "What's New" is much easier to discover in the app

We improved how product updates are announced and tracked:

• A "What's New" item in the top menu now shows an unread indicator when there are new updates.
• Important updates can appear as in-app banners/nudges with direct links.
• A dedicated "What's New" page now lists updates, highlights unread items, and supports direct links to a specific update.

Why this matters

• You get fewer low-confidence schedule suggestions.
• When you do get a suggestion, you can quickly see whether it is backed by enough real data.
• You are less likely to miss important product updates that affect your workflow.