AI Recommendations
What are AI Recommendations?
AI Recommendations is the synthesis layer that sits above all 11 other models. Rather than requiring you to check each model separately, it surfaces the most impactful actions for today across your entire customer base.
How recommendations are generated
The system runs every 2 hours. For each recommendation, it considers:
- Impact — How much revenue is at stake?
- Urgency — Is a win-back window closing? Is churn accelerating?
- Confidence — How certain is the underlying model?
- Actionability — Does LiftSignals have a connected platform to act through?
Recommendation types
| Type | Example | Source model |
|---|---|---|
| Churn spike alert | "212 customers entered Critical band in the last 48h — export to Klaviyo" | Churn Detection |
| Win-back window | "312 customers' optimal win-back window closes in 48h" | Win-back Timing |
| CLV tier opportunity | "148 Developing-tier customers are $1 away from the next tier" | CLV Forecast + LTV Maximizer |
| Repurchase opportunity | "890 customers are overdue for repurchase — export for nudge" | Repurchase Prediction |
| Data quality alert | "First_order_date missing for 4,200 records — reducing CLV accuracy" | Data health |
Where to find recommendations
Recommendations appear on the Today screen as action cards — each with a one-click action. The most urgent recommendations are shown first. You can also access the full recommendation list from the AI advisor panel in the topbar.
Was this article helpful?