The LLM positioning lag: What it is and how to avoid it
Businesses that reposition themselves are finding that AI assistants keep describing them the way they used to be. The gap between a company's current positioning and what a large language model says about it is being called the LLM positioning lag, and it can persist for years after a rebrand or a move into a new market.
The cause is largely structural. Models work from training data that may predate the change, and they only search the web when a question appears to need it or when a user turns search on. Where old and new descriptions of a company coexist online, the model has to choose which to trust, and a long record of the old positioning can outweigh a recent update.
Examples include Hotjar, which Contentsquare acquired in 2021 and merged with by July 2025, yet ChatGPT still lists as an independent heatmapping tool. Blue Apron dropped its subscription requirement in August 2025, but ChatGPT does not mention the pay-as-you-go option. Duolingo now offers maths, music and chess alongside languages, and ChatGPT still describes it as a language app.
For brands, the practical effect is that prospects asking an AI assistant for recommendations may never reach the company's own site, where the current positioning is stated. The share of consumers using generative AI instead of search engines for product and service recommendations rose from 25% in 2023 to 58% in 2025, according to Capgemini.
Reported by Semrush Blog ↗
Topics: ChatGPT, Citations, Measurement