How Content Scoring Works
Understanding how Radarpilot selects and ranks content helps you create templates that reach more users. The scoring system balances relevance, user preferences, and freshness to deliver the best content at the right time.
Pool Building
When content is needed, the system first assembles a pool of candidate templates from the active pack:
- Context match — Templates matching the requested context (e.g. calendar, email, task)
- General pool — General-purpose templates are always included as fallbacks
- Broad matches — Related contexts are included (e.g. a morning template can match general)
- Deduplication — Duplicate template IDs are removed
Scoring
Each template receives a composite score from 6 factors. Higher scores mean the template is more likely to be shown.
| Factor | Points | How it works |
|---|---|---|
| Context match | 0 – 3 | Exact context match = +3, broad match = +1.5, general fallback = +0.5 |
| Tone preference | 0 – 2 | Template tone matches user's preferred tones = +2 |
| Creator preference | -0.5 to +1 | Selected creator = +1, different creator = -0.5, no creator = 0 |
| Engagement | -0.75 to +1.25 | Based on average engagement score from the last 14 days, plus +0.5 if the user completed a task from it |
| Recency penalty | -2 | -2 if the template was shown in the user's last 10 interactions |
| Time of day | 0 – 1 | Morning templates get +1 in AM, evening templates get +1 in PM |
Selection
Templates are sorted by score descending, then selected using two methods:
Single template
Weighted random pick from the top 5 candidates. Higher-scored templates are more likely to be chosen, but there's always variety.
Multiple examples
Picks up to 4 templates — tone diversity first (one per tone), then fills remaining slots by score.
A note on creator filtering
Creator selection is a soft preference, not a hard filter. When a user selects a creator, their templates get +1.0 while other creators get -0.5. This means with the right combination of context, tone, and engagement scores, a template from another creator can still appear.
The server-side fetch does hard-filter by creator, but locally cached templates from previous fetches may still be mixed in to ensure the user always has content available.
Tips for creators
Use specific context tags. An exact context match scores 3x higher than a general fallback.
Align your tone. Templates matching a user's preferred tone get +2 points — a significant boost.
Create time-specific content. Morning and evening tagged templates get a +1 bonus at the right time.
Volume matters. With a -2 recency penalty, users need fresh templates. More content means more reach.
Drive engagement. Templates that users interact with and complete tasks from score higher over time.