How this is compiled

Sourcing

Records are researched from primary sources wherever one exists: the organiser's own event page, the organization's own charter or about page, the vendor's own pricing page, the company's own funding announcement, the platform's own documentation. Aggregators and round-up posts are used only todiscover candidates, never as the final source — several were found to carry wrong cities and wrong dates during compilation.

What "verified" means

A record is marked verified only when a page confirming its key facts was actually loaded during compilation. It does not mean a human attended the event or bought the software. Unverified records are still published — a real conference with an unconfirmable date is more useful than a silent gap — but they are flagged, and their detail pages say so.

Rules the compilation follows

• Never invent a URL. If a website could not be confirmed, the field is empty.
• Never estimate a number. Undisclosed funding, unpublished pricing and unknown attendance are left empty, not guessed. Totals shown on this site therefore understate reality.
• Never invent a date. Unannounced editions carry an explicit note instead of a plausible-looking date.
• Record the conflict. Where two sources disagree, the more authoritative one is used and the discrepancy is noted rather than silently resolved.
• Prefer fewer, correct records over padded lists. A short honest dataset beats a long fictional one.
• Public professional information only for people. No personal contact details, no private information, and removal on request.

Classification

Categories are descriptive and deliberately coarse. An event tagged geo is one whose programme substantially covers generative search — not merely one that mentions AI in a keynote. A tool's primary category is where most users would look for it first; secondary categories capture the rest. None of these labels imply quality, and none of them are rankings.

Known gaps

Compilation surfaced several honest weaknesses, and it is better to state them than to imply completeness:

• Regional coverage is uneven. Latin America, Africa and parts of East Asia are thinner than Europe and North America — partly because fewer events publish in English, partly because several national bodies' sites could not be reached during compilation.
• GEO is young. Only about a dozen events are exclusively about generative engine optimization, most are small, and a large share are run by vendors in the category. Very few 2027 dates are announced. The GEO events page says this plainly rather than padding.
• Pricing decays fastest. Vendors change prices without notice, and several block automated access to their pricing pages entirely, so those fields are empty rather than stale.
• Funding is understated. Only amounts stated in a primary source are recorded, and rounds announced in currencies other than dollars are noted in prose rather than converted.
• People data ages fast. Roles and employers change constantly. Where a current employer could not be confirmed, the role is phrased generically rather than guessed.

Corrections

Corrections take priority over new additions. If a fact here is wrong, saying so is the single most useful contribution you can make — and it costs nothing.

Submit a correction