GETMINT RESEARCH
GetMint has its own research team. They dig through the answers we collect every day, measure whether our recommendations actually move a brand’s visibility, and turn what holds up into the scientific method behind the product.
Illustrative: 100 sources cited today, re-measured day after day.
Product
Research shapes the product.
A feature ships when a study shows the behaviour behind it is real, and stable enough to act on.
Advice
Research backs every recommendation.
What we tell you to do traces back to a measured result, not to what the industry repeats.
Papers
Research gets published.
Method, scope and results go out as papers anyone can check, including what we couldn’t measure.
HOW WE MEASURE
AI answers vary every day and models change without notice. One question on one day tells you almost nothing. So we measure the way a statistician would.
1 question, 1 day
±28 points
About 20 questions per topic, over 14 days
±7 points
Precision of a visibility score, from GetMint research.
By topic, not by prompt
About 20 questions per topic, so one odd answer doesn’t move the score.
Over time
14 days of answers, to smooth out daily variation.
With a margin
Every score shows how precise it is, so you know when a change is real.
Model updates filtered
We separate what your actions moved from what a model update reshuffled.
WHAT THE DATA SHOWS
220,000+
domains analysed
49,000+
prompt groups across 545 brands
14%
of sources persist over time
63%
of sources are volatile
LATEST PAPERS
SOURCES · [DATE]
The Core Sources study: what AI answers keep relying on
220,000+ domains, 545 brands. Which sources persist, which disappear, and what it means for your content plan.
Read the paper →
MEASUREMENT · [DATE]
How precise is an AI visibility score?
From ±28 points on a single question to ±7 points on a topic measured over 14 days.
Read the paper →
CASE ANALYSIS · [DATE]
FAQ pages and ChatGPT: from 17% to 65% of answers
An education brand published FAQ pages on its target questions. ChatGPT started citing them, and the brand held the gain for weeks.
Read the analysis →
WHAT WE’RE STUDYING NOW
The questions on our research bench right now. Each one becomes a paper, whatever the answer.
01
[How long does a new page take to be cited by AI engines?]
02
[Do ChatGPT and AI Overviews rely on the same sources?]
03
[How much does a model update reshuffle a category?]
04
[Do ChatGPT Ads change which brands are mentioned in organic answers?]
THE RESEARCH TEAM
GetMint’s fundamental research team is led by a PhD in applied mathematics, former researcher at École Polytechnique. It studies how AI models work and which actions actually change a brand’s visibility.

LLM RESEARCH LEAD
Amine Aziz-Alaoui
Leads GetMint Research: how AI models build their answers, and which actions actually move a brand’s visibility.

CTO AND CO-FOUNDER
Emmanuel Costa
Turns research results into the product. Former CTO of Bankin’ and Bridge. ESSEC, Stanford GSB.