GEO for IPO Issuers: Why Your S-1 Should Read for Claude

Generative Engine Optimization, or GEO, is the new IR website project. FGS Global named it as one of the four pillars of its AI Advisory practice. Brunswick built an entire offering around it called Algorithmic Relations. The idea is simple: more of the people researching your company are now asking large language models, not search engines. If your story is not legible to those models, you are losing the first impression to whatever the model decides to surface.

What changes when an analyst asks Claude instead of Bloomberg

Three things. First, the model summarizes from its training corpus and any retrieval source it has access to, which means your IR site, your latest 10-K, and your most recent earnings call may or may not be in the answer depending on how recently the model was indexed. Second, the model paraphrases. Word choice that survives a careful read by a portfolio manager may not survive a model summarizing it for a buy-side analyst. Third, the model cites. If your S-1 risk-factor language is repeated verbatim in an analyst's research note via a model, that language is now de facto your disclosure on the topic.

Four moves that matter on the IPO timeline

1. Run a baseline GEO audit pre-filing. Ask the four major models the same ten investor questions about your sector and your company. Catalog where the answers are blank, generic, or wrong. That is the gap your S-1 has to close.
2. Optimize the IR website for retrieval. Structured data, semantic markup, well-named anchors, and machine-readable fact pages dramatically improve the chance that a model retrieves your content rather than a third party's summary.
3. Write your risk factors and MD&A for both audiences. The lawyer reads it for liability protection. The model reads it for facts. Both have to walk away with the same picture.
4. Monitor and remediate. After the IPO, set up monitoring across the major models. When a model starts answering a key question wrong, you have a content gap, not a model bug. Fix the gap.

The cost of doing nothing

FGS and Brunswick are charging real money for GEO work today. The reason boards are buying it: a single mis-summarized risk factor in a Perplexity answer can travel to fifty buy-side analysts before the IR team even sees it. The IPO window does not give you time to clean that up after the fact.

ICR Capital and the AI & Intelligence practice work together on this. Capital handles the S-1 narrative and bank selection. AI & Intelligence handles the GEO baseline, the IR-site retrieval architecture, and the post-pricing monitoring. The two are now one workflow.