How we score an account.

Every analysis produces a single risk score from 0 to 100, where lower is better. The score is a weighted composite of individual signals, each calibrated against patterns observed across genuine and fraudulent accounts. It maps to four plain-language tiers:

The IREM framework — our Influencer Risk Evaluation Model — organizes the assessment into five dimensions, so a single number is always accompanied by a structured view of where risk does or does not concentrate.

Each platform has its own analyzer with platform-specific signal categories, weighted to reflect how strongly each one predicts inauthentic audiences on that platform. The headline weights are below; within each category we evaluate dozens of individual signals.

Weights are tuned over time as we calibrate against accounts with known outcomes, so the exact figures may shift. The structure — and the principle that no single signal decides a verdict — does not.

Alongside the score, every report carries a confidence level. Confidence reflects how strongly the signals agree with one another: high confidence means multiple independent indicators point to the same conclusion; lower confidence means the data is thinner or the signals are mixed, and the verdict should be weighed accordingly.

This analysis is generated from publicly accessible data at a single point in time. It is an analytical aid — signals and estimates, not accusations or a professional fraud investigation or legal determination — and should be used as one input alongside your own due diligence.

Some metrics are estimated where platforms restrict access to underlying data. For example, follower-sample quality or comment text may be unavailable for certain accounts, in which case Veriscore infers quality from engagement ratios and notes the estimation. Estimated figures are inherently less precise than directly measured ones.

Audience and engagement metrics change over time. For high-value partnerships, refresh the analysis close to the decision date and request the creator's first-party platform analytics to corroborate these findings. If you believe a result is inaccurate, you can re-run it with fresh data or request a correction at romil@regnor.systems.