AI Insights Panel

Understanding the automated analysis metrics

4 min read SPF Management

AI Insights Panel

The AI Insights panel provides a real-time snapshot of your domain's email source health, powered by analysis of your incoming DMARC aggregate reports. It sits at the top of the SPF Management page and gives you the headline numbers before you dive into individual sources.

The Metrics Explained

Metric Target What It Means
Average Confidence 70%+ The weighted average AI trust score across all pending sources. Higher means your sources are well-understood and likely legitimate.
Sources Analyzed Varies Total number of distinct email sources (IPs or services) detected in DMARC reports and waiting for review.
Ready to Approve As high as possible Sources the AI recommends adding to SPF right now. High confidence + strong authentication track record.
Anomalies Detected 0 Sources showing unusual behaviour — sudden volume spikes, new IPs from suspicious ranges, or authentication drops. Investigate these first.
SPF Lookup Count < 10 Current number of DNS lookups in your managed SPF record. The hard limit is 10 — exceeding it causes SPF to fail for all email.

How Confidence Scores Work

Each source receives a confidence score (0–100%) calculated from multiple signals:

Increases Score

  • Recognised email service (SendGrid, Mailchimp, etc.)
  • High SPF + DKIM authentication rates (>95%)
  • Consistent sending history (14+ days)
  • DKIM alignment present
  • Volume proportional to domain's normal patterns

Decreases Score

  • Unknown or suspicious IP ranges
  • Zero or very low authentication rates
  • Sudden volume spikes (>200% increase)
  • No DKIM alignment
  • Residential or shared hosting IPs

What Zero Values Mean

  • 0 Sources Analyzed — No DMARC reports received yet, or no new pending sources. Normal for domains configured within the last 24–48 hours.
  • 0 Anomalies — All sources are behaving within expected patterns. This is the healthy state.
  • 0 Ready to Approve — No high-confidence pending sources. Either all sources are already approved, or existing pending sources need more data before a recommendation can be made.

Tip: Use the Anomaly Count as Your Priority Signal

If anomalies are greater than 0, investigate those sources before approving anything else. Anomalies may indicate a compromised account or spoofing attempt.