← All terms
core

Model Drift

Statistically significant changes in an AI model's citation behaviour over time following training updates – brands that were frequently cited can suddenly disappear from AI responses without warning.

Model Drift is a statistically significant shift in which sources an AI model cites, driven by training or algorithm updates rather than by any change to your own content. A brand cited in 70% of AI answers can fall to 35% after a single model update — so the same prompt set must be re-probed over time to catch it.

This term is part of the 3LA EntityMap → /entitymap.html

Related terms

FROM THE BLOG

Latest field notes.

View all posts
Entity Accuracy Engine - Why we built it and who it's for
Insights

Entity Accuracy Engine - Why we built it and who it's for

LLMs hallucinate about companies, prices, key people and mandates - and it's costing NGOs, universities, research bodies, financial institutions, political actors and agencies real leads and real trust. Here's the background of the Entity Accuracy Engine, and exactly who I built it for.

Aug 24, 2026
AI SEO Tools Comparison in 2026: A Strategist's Honest Take
SEO

AI SEO Tools Comparison in 2026: A Strategist's Honest Take

Classical AI SEO tools grade your content with an LLM. The Three-Layer Approach measures - deterministically - whether AI can find, trust and cite you. Here's the honest comparison. This article was first published on nettpilot.com

Jul 5, 2026
Your GEO Strategy Is Failing Three Out of Four AI Audiences
Insights

Your GEO Strategy Is Failing Three Out of Four AI Audiences

ChatGPT, Claude, Perplexity and Gemini do not share a user base. They share a category name. A 2026 study found only ~11% of domains cited by ChatGPT are also cited by Perplexity. Here is the field guide to a platform-segmented GEO strategy.

May 3, 2026