What is AI Score?

AI Score is a metric in the ClickPatrol dashboard that summarizes how likely a click is to be non-human or otherwise invalid, based on our analysis of that click and its context. You may also hear it called a fraud score. It is not a public Google metric; it is our consolidated signal to help you prioritize review and understand risk levels across traffic.

How we assign AI Score

We analyze each click that we observe for your protected campaigns using multiple data sources and models. Signals can include IP and network reputation, device and session behavior, patterns consistent with suspicious behavior, and how the click compares to typical traffic in your vertical and region. We combine those inputs into a single score band so you can scan activity quickly without reading every raw field.

What the bands mean

We group clicks into five categories from safest to highest risk:

• Safe
• Low
• Somewhat medium
• Medium
• High

Clicks in medium and high bands are treated as suspicious clicks for reporting purposes. That does not mean we block every medium or high score automatically. Blocking is stricter: we act when we are confident there is no genuine human value behind the interaction, and we also account for your account’s vertical and legitimate edge cases (for example, some B2B advertisers receive real prospects from data center or hosting IPs that would look risky by default).

How you should use AI Score day to day

Use AI Score as a compass, not the whole map. A higher score means “investigate or trust our automated handling,” while safe and low scores are consistent with normal traffic. Pair the score with time ranges, campaigns, and predicted clicks saved to see whether protection is aligned with your tolerance for risk versus conversions.

Our default settings aim to avoid blocking traffic that might still convert. If you tighten blocking in the dashboard, you may reduce waste further but increase the chance of excluding borderline human traffic. That tradeoff is why we expose the score and let teams tune policy where appropriate.