# How to interpret Wayfold analytics

Sessions, estimated daily reach, AI traffic and before/after comparisons: definitions and measurement limits.

Source: https://wayfold.ch/en/methodology

Updated: 2026-10-20

## Sessions are not people

A Session groups activity into a short-lived anonymous journey, with an absolute eight-hour limit. Separate sessions are never linked into a returning-person profile. Session counts must not be presented as unique people.

## Estimated Daily Unique Visitors

Daily reach is a separate estimate. Counts use daily rotating secrets and HyperLogLog registers; they do not store a list of visitors. The 1.625% standard error describes the estimator, not the accuracy of identifying people on shared or changing connections. Do not sum daily values and call them unique monthly visitors.

## AI traffic has different meanings

Recognizable referrals from AI assistants describe visits arriving from those services. Classified crawler or retrieval requests describe machine access. Neither proves that a brand was cited in an answer, and missing referrers or unobserved requests mean coverage is incomplete.

## Comparisons and external sources

A before/after change is an association, not proof of causality. Sample size, seasonality and simultaneous changes affect interpretation. Search Console and imported GA4 data remain separately labelled; their counts are not silently added to Wayfold counts.

## Data minimization and limits

Analytics does not store IP addresses or full user agents. Server-side classification may use them transiently. Input filtering reduces accidental personal data but cannot guarantee that arbitrary customer-authored content is anonymous. Session-linked detail is retained for 365 days from last receipt; count-only rollups for 730 days, subject to the documented retention process and access safeguards.