Skip to main content
Guides

SearchBastion Traffic Quality Score: Interpret the Rules

SearchBastion Traffic Quality Score: Interpret the Rules

SearchBastion's tracked-click quality score is a rules-based diagnostic from 0 to 100. Higher values mean fewer deductions under those rules. It is not a probability of being human, an advertising quality score, or a guarantee that a lead is valuable.

What the tracked-click calculation evaluates

The current implementation starts at 100 and applies deductions for signals including user-agent patterns, selected referrer patterns, configured country rules, IP classification and cross-site IP reputation. The result is clamped to the 0–100 range. Duplicate and velocity flags are separate observations; do not assume every flag directly changes this particular calculation.

An example from the implemented rules

If the only applicable deduction is the tracked click's VPN flag, the current rule subtracts 30, giving a score of 70. If other rules apply, the result can be lower. This is an arithmetic illustration of the implementation, not evidence that 30 percent of the visit is fraudulent. Rules may change, so use the deployed rule set when reproducing a historical result.

The public IP checker uses a different rule set. Two scores for the same address can differ without either being a count of verified bots.

Investigate a low score

  1. Read the individual flags and context instead of sorting only by the total.
  2. Check the relevant tracked event and whether a valid action was completed.
  3. Compare repeated patterns over the same campaign and time window.
  4. Review false positives from corporate networks, privacy services and legitimate automated tools.
  5. Test a narrowly scoped response before exporting or applying a blocklist.

Measure whether the score is useful

Periodically compare flagged interactions with independently reviewed outcomes such as accepted leads. Keep ambiguous outcomes separate. Do not treat the same heuristic as both the prediction and the evidence that the prediction was correct. A risk threshold can help prioritize work, but it cannot certify fraud, quantify confirmed financial loss or prove a network should be blocked.

Inspect network context alongside your access-controlled click records. This description is based on SearchBastion's deployed scoring implementation, not an externally certified detection benchmark.

Advertisement
#Click Fraud #Quality Score

Check your website for free

Run an SEO audit or check your traffic quality now — no signup required.