Loading...
As search moves inside ChatGPT, Perplexity, and Google AI Overviews, agencies are selling a new flavor of magic. Here is the checklist of what to avoid—and what actually gets your business cited.
Generative Engine Optimisation (GEO) is the next generation of search visibility. But just like the early days of SEO, the gap between what works and what is sold is widening.
The mechanism AI search engines use to recommend a business is simple: retrieval and citation. They read the web, extract facts, evaluate sources for substance and authority, and cite them.
Because this process is algorithmic, some agencies claim they can "hack" or "train" ChatGPT to recommend you. But manipulative shortcuts don't work on systems trained to detect low-quality, generic text.
If an agency or software vendor promises any of the following, tread carefully.
ChatGPT and Perplexity synthesize answers dynamically for every prompt. Rankings do not exist in a stable list format anymore. Promising a 'No. 1 spot' in an AI summary is a sign of traditional rank-tracking vanity.
There is no 'submission portal' for ChatGPT, Claude, or Gemini. AI search assistants find your business by crawling the public web in real-time. If you block their search bots (like OAI-SearchBot), no submission can help.
JSON-LD schemas are excellent for structuring data (like FAQs or prices), but they do not brainwash LLMs. AI search engines cite sites because they have substantive, authored content with clear evidence—not because a hidden tag claims they are the best.
LLMs are trained to detect and ignore low-quality link spam and directories. Creating hundreds of low-authority directory listings provides zero benefit for AI citation and reads like typical web-era link-building schemes.
An llms.txt file is a useful map for LLMs to read your site structure, but it is not a magic visibility shield. Without ongoing, high-quality content updates, a static file becomes stale and useless in weeks.
Manipulative tactics do more than fail—they carry severe domain-level risks. Large Language Models are built on top of web search indexes (Bing and Google). If your domain is flagged for search spam, it will be ignored by both web crawlers and AI answer engines.
Fake reviews or reviews bought in bulk to manipulate sentiment analysis.
Astroturfing: Generating fake Reddit, Quora, or forum conversations to force consensus.
Mass-generating thousands of thin AI-written blog posts or answer pages.
Wikipedia or Wikidata manipulation to fabricate entity history.
Cloaking content (showing search bots different content than users see).
AI search visibility is built on domain trust. manipulative strategies risk permanent blacklisting from Bing index and Google search.
Real AI search visibility is built the same way as real authority: through technical legibility, structured data, and substantive content.
Checking robots.txt to ensure your site does not block search agents like OAI-SearchBot or PerplexityBot. If the bots are blocked, you will never be cited.
Ensuring your business Name, Address, and Phone (NAP) and core facts are matching across your site, local profiles, and directories so LLMs can associate them correctly.
Publishing structured, answer-first content that directly responds to the prompts and questions buyers ask search engines.
Weaving concrete metrics, specific case studies, and clear authorship credentials into your writing. Generic AI text is what AI engines filter out.
Encouraging genuine reviews on third-party trust sites, which LLMs scrape for sentiment analysis and ratings recommendations.
Ghostart is the anti-beige platform for GEO. We don't sell ranking hacks. We help you and your team publish the high-substance, citable content that AI actually recommends.