# AI referral conversion and engagement: methodology

Updated: 8 August 2026

## Scope

This package presents public AI referral benchmarks from Similarweb and Adobe Digital Insights. The two datasets are kept separate because their coverage, periods, metrics, and comparison groups differ. No proprietary BGraph or customer data is included.

## Source inputs

- Similarweb: ecommerce conversion estimates for AI referrals, paid search, and organic search in June 2025.
- Adobe Digital Insights: AI referral traffic growth and engagement differences for retail visits from July 2024 through February 2025.

## Transformations

Published values were transcribed into rows with an explicit source and period. BGraph calculated a 22.6% conversion lift over paid search using 11.4 divided by 9.3 minus 1. BGraph calculated a 2.15 multiple versus organic search using 11.4 divided by 5.3.

## Reproducibility

Filter `source_name` before comparing rows. Similarweb conversion rates can be compared within their June 2025 sample. Adobe engagement and conversion differences can be compared within Adobe's retail analysis. Do not pool the two publishers into one conversion or engagement estimate.

## Limitations

- Both publishers use proprietary traffic measurement methods and estimated metrics.
- A conversion-rate difference does not identify incremental revenue or causal lift.
- A high rate can coexist with low referral volume.
- Device mix, category mix, geography, attribution window, and conversion definitions may differ.
- The public extracts do not provide sampling uncertainty or raw observations.

## Preferred citation

BGraph Research. (2026). AI referrals can be small in volume and strong in intent. Updated 8 August 2026. https://bgraph.io/research/ai-referral-conversion-engagement.