ELI RESEARCH
Eli AI Search Benchmark data
Research is useful when readers can inspect data behind a chart.
No public benchmark release yet. The schema and methodology are public; first release follows completed, normalized, quality-checked observations.
RELEASE STANDARD
Methodology is publishable before findings are.
Eli can publish the question universe, schema, definitions and quality controls before a benchmark release. It will only publish an aggregate when the underlying observations are complete, reviewable and described with their coverage and limitations. Until then, this page makes no performance, ranking or customer-outcome claim.
Dataset schema
Required fields include observation_id, prompt_id, intent, variant, engine, model, date, market, language, raw answer reference, brand, mentioned/recommended, context, citation URL/domain, source type and notes.
Release metadata
Every release includes release ID, methodology/prompt-universe version, field dates, observation count, engines, languages, markets, exclusions and correction history.
Download formats and raw answers
When generated, offer CSV, JSON and human table. If raw answers cannot be redistributed, publish metadata, classifications, citations, hash/reference and limited excerpts where appropriate.
Corrections
Issue corrected release, preserve prior version where appropriate and document change; do not silently rewrite history.
FAQ
Questions about the research design and release boundary.
Is dataset live?
Only after a verified release.
Can I use data?
A reuse/license statement will accompany first release.
Raw answers included?
Only where technically and legally appropriate.
How corrections?
Versioned release notes.
Can metrics be reproduced?
Schema/methodology aim to allow aggregate reproduction.
CONNECT YOUR WEBSITE
Inspect the dataset schema
This page is the data layer behind Eli's AI-search research. Each public release should expose field schema, benchmark version, field dates, prompt metadata and downloadable tables needed to reproduce published aggregates.
Inspect the dataset schema