Smart Policy Recommender

​

Overview

The Smart Policy Recommender analyzes your interaction history data to identify contact frequency patterns and recommend optimal contact policies. It looks at how often customers are contacted, through which channels, and how they respond — then suggests policies that reduce fatigue and improve engagement. Navigate to AI > Policy Recommendations in the sidebar to view the dashboard, or trigger analysis programmatically via POST /api/v1/ai/analyze/{type} with type=policies in the path. The dashboard page (/ai/policy-recommendations) runs two intelligence tools automatically on load and every 5 minutes:

• analyzePolicyConflicts — Detects contradictions, overlaps, and misconfigurations between your policies
• analyzeContactPolicySuppression — Identifies channels or policies with abnormally high suppression rates

​

What It Analyzes

The recommender examines two dimensions of your interaction data:

• Frequency bands — How contact frequency (contacts per week) correlates with conversion rates. It identifies the “sweet spot” where customers are contacted often enough to stay engaged but not so often that they disengage. The system generates frequency bands at 1, 3, 5, 7, and 10 contacts per week.
• Channel performance — Which channels produce the best engagement. For each channel, the analyzer tracks impressions, clicks, and conversions to compute click-through rate (CTR) and conversion rate (CVR).

​

Analysis Output Format

Each analysis returns one or more policy-recommendation objects with this structure:

​

Dual-Tier Analysis

The recommender supports two analysis tiers, automatically routed by the platform’s ML router: The ML Worker tier produces more precise recommendations because it processes the full dataset rather than a summary. KaireonAI automatically routes to the ML Worker when it is available and the dataset is large enough to benefit from it. If the ML Worker submission fails, the system falls back to LLM analysis.

​

Heuristic Fallback

If the LLM call fails (provider error, timeout, etc.), the recommender falls back to a deterministic heuristic analysis that:

1. Finds the optimal frequency — Identifies the contacts-per-week band with the highest conversion rate
2. Detects drop-off — Finds where conversion rate drops below 50% of the peak
3. Recommends a frequency cap — Sets the cap one step below the drop-off point
4. Flags low-engagement channels — Channels with CTR below 2% and at least 100 impressions get a “reduce frequency” recommendation
5. Highlights high-converting channels — Channels with CVR above 10% and at least 50 clicks get a “prioritize” recommendation

​

Reading the Frequency Charts

The recommender displays an interactive frequency analysis chart:

• X-axis — Contact frequency (contacts per week)
• Y-axis — Conversion rate (percentage)
• Curve — Shows how conversion rate changes as contact frequency increases
• Optimal zone — Highlighted region where the frequency-to-response ratio is highest
• Current average — A marker showing your current average contact frequency

The chart data comes from the metadata.frequencyBands array in the recommendation response.

​

Applying Recommendations

When you click Apply on a recommendation:

1. The Apply button arms on first click (“Confirm apply?”) and executes on the second (admin only)
2. A new Contact Policy is created in draft status, pre-configured with the recommended frequency limit, time window, and channel scope
3. The recommendation is marked Applied with the created policy’s id; review and activate the draft on the Contact Policies page

Run the analysis from AI > Policy Recommendations with the Analyze policies button — proposals land in the inbox on the same page (deduplicated against open recommendations).

​

Best Practices

• Run analysis regularly — Customer behavior changes over time. Re-run the recommender monthly or after major campaign changes.
• Start with global policies — Apply broad frequency caps first, then refine per-channel.
• Compare tiers — If both LLM and ML Worker are available, compare their recommendations. The ML Worker is more precise for large datasets; the LLM can surface qualitative insights the ML Worker misses.

​

Advanced Parameters

Each policy analysis run can be fine-tuned using the Advanced Parameters panel. Expand it to adjust: Per-run overrides apply only to that analysis and do not change your saved tenant configuration. To change organization-wide defaults, go to AI Configuration.

​

Large Dataset Warning

When the interaction history contains 5,000 or more rows, a confirmation dialog appears before analysis begins. The dialog shows:

• Accuracy comparison — ML Worker uses logistic regression and statistical analysis on the full dataset vs. LLM pattern matching on summarized data
• Estimated cost — Token count and approximate cost if proceeding with LLM
• Speed comparison — ML Worker processes locally in seconds vs. LLM round-trip

You can choose Use ML Worker (recommended for large datasets) or Proceed with LLM (uses summarized data).

​

Next Steps