We measure our own portfolio — in public.
CiteGraph is a measurement company, so we publish our measurements. Below are real API & developer-tool sites scored on AI Citation Share — the share of a category's buyer prompts where the site holds a linked citation in an AI answer.
oilpriceapi.com — the benchmark (~87%)
A strong, established API. Cited in roughly 87% of oil/commodity-price-API prompts across both engines we tested — this is what "good" looks like. Its citations come from a mix of authority sources (eia.gov, data.gov), its own docs, and community sources (GitHub, Reddit). Strong domains are cited at the domain level almost regardless of the specific page, so for these the work shifts to which page wins the click and recommendation rank.
machineshop.directory — mid-authority (22%)
Cited 22% of the time for CNC-machine-shop-directory
prompts. It wins the head term ("directory of CNC machine shops") but is
invisible on the long tail: real, indexed pages
(/machine-shops/texas, /certifications/as9100)
that exist but win 0 citations. That "exists ≠ cited"
gap is exactly what CiteGraph surfaces.
foodchainapi.com — newer API, mostly invisible (8%)
An FDA Food Traceability List / FSMA-204 API. Despite already shipping some structured data and a real content blog, it's cited only 8% of the time across food-traceability prompts. It wins its single most on-the-nose query ("FDA Food Traceability List API" — cited in 60% of runs) but is invisible on the rest of the category (FSMA-204 compliance, traceability software, the developer queries). Schema is necessary but not sufficient — authority and earned citations matter too. Lots of headroom.
gettariff.com — a live lift experiment (0% → tracking)
This is the one to watch. gettariff has 21 content pages covering its whole category — yet measured 0% Citation Share on June 19, because 0 of those 21 pages had any structured data. Content-rich, schema-blind, AI-invisible.
CiteGraph diagnosed the gap and we shipped the fix — JSON-LD (Organization, WebSite, WebApplication) across the site, live in production June 19. Baseline is locked at 0%. We're re-measuring at +1, +2, and +4 weeks as the answer engines reindex, and we'll post the lift chart here. No claims until the data lands — that's the whole point of measuring.