
Post originally published on Dzianis’ Medium blog
More and more businesses are taking AI visibility seriously and use services that help them appear in ChatGPT/Claude/Copilot/Gemini recommendations.
However, even if you manage to get mentioned by AI tools for certain queries, it’s too early to relax. At WebCopyLand we’ve been seeing a lot of fluctuation with AI visibility — because of the very nature of AI tools.
One week you may be getting clicks and sales from AI tools, then it can all go dry, then reappear again.
Here’s why that happens:
LLMs are probabilistic
Depending on the prompt, even the minor accents, an LLM may consider your page a good source to cite — or it may not. With SEO the results would be ranked for higher-volume queries, updating relatively slowly, and we’d all be fighting for those spots.
LLM usage is conversational: consumers chat rather than outright search. As a result each chat history and nuance affects the suggestions that the LLM deems the most statistically appropriate.
The logic is different from LLM to LLM
This one is self-explanatory: you may come up in cited sources on ChatGPT but stay invisible on Claude.
When we run AI visibility audits @ WebCopyLand we build lists of things highly visible brands do differently on each AI platform, and you should too.
AI Tools are biased
This bias comes from both the training data and “factory settings”, so your visibility will depend on how well you fit the bias. All kinds of censorship and safety filters affect the sources they cite, all of that changes with time and changes your visibility.
LLMs get re-trained
As the tech evolves, training data updates happen more and more often. Years ago ChatGPT knew nothing about news and market updates, and only had the conversational level of an average redditor — it’s much more advanced and up-to-date now.
Training data is generally a synonym for “the Internet”. LLMs index the web top to bottom, especially online communities. The closer your online mentions are to real human communities and authority sources, the
So this means that AI visibility is competitive, just like SEO. If other brands do something right within the next training data batch (i.e. online)
AI Models get updated
Not just the training data, but the way they build answers and the logic variations they use. A “smarter” model may suddenly stop recommending your brand, or vice versa — start mentioning it.
LLMs get personalized
Personalization is the zeitgeist of the web. Depending on the user’s browsing history, purchase history, location, and even network, they will eventually be getting different responses from LLMs.
All of this means that it’s extremely hard to plan or predict any results for your AI visibility campaigns. Even less than with SEO, which was already filled with variables.
AI visibility is unstable by design, and it will remain so either until the market gets monopolized (like SEO was by Google) or until the whole thing crashes. In any case, you want to be on board.
The best thing you can do it just keep doing the right things and stay in the game.
I’ve posted a list of generative AI search ranking factors before, let me recap in a quick AI visibility checklist:
- constantly publish content: specific, detailed, conversational, centered around your offer and pain points of your audience. Try to come up with topics that resemble questions your target market sends into AI tools;
- be visible online: run frequent press releases, social media campaigns, try to go viral, get mentions where your target market is;
- become an authority by a) sharing authoritative thought leadership content based on unique industry experience and b) being mentioned on the top websites in your niche;
- have a content distribution system — scale and repurpose every bit of content you produce;
- be fast & live in the future: hijack news, give interviews, predict trends;
- get creative with content types — produce books, industry reports, start a podcast, and so on.
It’s a lot of work but there’s no way around it.
You can buy ads, but you might have a hard time outbidding organically visible competitors.
Need help with your AI visibility strategy?