Perception Drift
The week-over-week change in how AI models describe a brand, driven by shifts in the model's source mix.
Definition
Perception drift is the measurable change, week to week, in the wording AI models use to describe a brand. The brand may be unchanged; the model's retrieved source mix is what shifted.
Why it matters
Drift compounds. A small weekly shift in framing turns into a quarter-long repositioning that nobody on the brand team ratified. Catching drift early is the difference between a fix and a rebrand.
How to track it
Run a fixed prompt set on the leading AI models every week, score each response across positioning, sentiment, target-audience, and competitor-proximity metrics, and alert on week-over-week deltas above threshold.
See how your brand performs across AI assistants
Strajist tracks your visibility, share of voice, and citations across ChatGPT, Gemini, Claude, Perplexity, and more.
Start free trialRelated terms
The discipline of measuring how AI models describe, categorize, and contextualize a brand, and which sources shape those descriptions. One layer beneath visibility tracking.
Tracing each AI description of a brand back to the specific URLs, press articles, reviews, and structured sources that taught the model.
How AI models rank and frame a brand against a defined competitor set on identical prompts, order, leader vs alternative labels, and recommendation strength.
How large language models describe, categorize, and contextualize a brand when asked about it, effectively the brand image AI models hold.
Whether AI-generated answers describe a brand in positive, neutral, or negative terms.
Tracking which competitors AI models recommend in your category and how their visibility evolves over time.