Methodology

What OnlineSummit indexes

OnlineSummit turns official summit content into a structured, searchable database. The unit of our work is not the event listing. It is the talk: who spoke, in what role at the time, at which summit, about which topics, and what they actually claimed.

For every talk we aim to record:

1. The talk itself: title, session type, date, duration and language.
2. The speaker as they were then: name, and the title and organization they held at the time of the talk, even if they have since moved on.
3. Structured content: an original executive summary, key takeaways, and atomic claims (facts, opinions, forecasts, statistics, company claims) each tied to a location in the source.
4. The source: the official video, agenda, slides, transcript or PDF the content came from, with timestamps or page numbers.

Where our data comes from

We index publicly available, official material: summit websites and agendas, official video channels, published slides and reports, organizer press releases, and speakers' own published talk materials. We do not bypass logins, paywalls or access controls, and we do not republish full videos, decks or transcripts. Our model is to index, summarize, structure and cite, not mirror.

Verification and review states

Every record carries its processing state. Three matter most to readers:

• Source Verified: the primary source has been located and checked.
• AI Summarized: summaries, takeaways and claims were machine-drafted from the source.
• Editorially Reviewed: a human editor has reviewed the record.
• Organizer Verified - an authorized organizer representative has confirmed the event-level material or relationship. This does not replace editorial review.

Machine-drafted content follows strict rules: no fact enters a published page without a source; exact quotes are only published when they can be checked verbatim against a transcript or clear audio. Otherwise we publish clearly labeled paraphrases. A speaker's opinion is never rewritten as industry fact, and promotional statements are labeled as company claims.

Quality thresholds

Pages that do not yet meet our content thresholds (for example, a speaker page with only one unverified talk) stay out of search-engine indexes until they do. We would rather have fewer, deeper pages than many thin ones.

Editorial sample data

No fictional records are published today. Every summit, talk, speaker and organization on this site comes from a real event with a verifiable source.

The rule exists because samples were used during product validation, and it stands if we ever need them again. A sample record is fictional, its record ID begins with sample:, its source status is sample, and it is excluded from search indexes. Samples never use real company logos or attribute invented statements to real people or organizations, and one may be replaced only by a separately verified record — never silently converted into evidence.

Ranking principles

Search and featured placement use relevance, verification status, recency and explicit editorial selection. Payment, sponsorship and organizer participation do not buy ranking. Promotional statements remain labeled as company claims even when they come from an official source.

Corrections

Use the report page for a misattributed quote, outdated affiliation, broken source, identity correction or rights request. We acknowledge complete requests within 3 business days, normally resolve straightforward corrections within 7 business days, and review credible urgent rights requests as soon as possible.