Methodology
Understand Mentionloom’s brand states, recommendation share, sampling protocol, evidence records, and the limits of API-based measurement.
Explore page
Tools & Research
AI Recommendation Index
Mentionloom’s AI Recommendation Index is a planned research benchmark. We are publishing the measurement design before publishing rankings so future results can be audited instead of accepted as a black box.
Get a snapshotA closer look at the measurement.
MethodologyPublished
Question setTo be disclosed
Collection windowPending first release
Brand rankingsNot published yet
What will the index measure?
Recommendation Share by category, buyer-intent stage and declared AI surface; Lost Recommendation Share; recurring citation patterns; engine divergence; and answer volatility across repeated samples.
What will it not do?
It will not mix recommendation visibility, referral traffic and crawler activity into one score. It will not present a single API sample as representative of every consumer answer, and it will not claim a cited source caused a model’s recommendation.
What is published today?
The methodology is published; the benchmark dataset is not. No leaderboard, “most recommended” brand or category ranking on this page is a measured result yet. When the first index is released, the sample window, surfaces, question set design and uncertainty method will accompany it.
Questions people ask
Why publish the methodology first?
So the definitions and limitations are fixed before there is an incentive to shape them around an interesting result.
Will the index rank brands from one answer?
No. The design uses repeated sampling across a defined question and engine set, with the collection period and uncertainty disclosed.
EARLY ACCESS
Start with your website.
Request a recommendation snapshot built around your company and likely buyer questions.