Proof, not promises.
Every Monday our automated tracker runs target prompts against Perplexity and Gemini, and we log ChatGPT and Claude citations by hand until their APIs expose them, capturing any citation that names Capture That Media or our client domains, time-stamped and posted here. No agency in San Antonio runs this in the open.
276
Automated checks logged
37
Target prompts tracked
0
Live citations captured · 2 engines automated
Monday
8am CT · weekly cron
Active tracking · no sample rows
Engines. Prompts. Cited passages.
We do not show placeholder or sample passages here. When a real citation is captured, it appears on this wall.
Tracking — first capture pending
No live citations posted yet.
You’re looking at a live tracker, not a brochure. We’ve run 276 automated checks across 37 target prompts and have 0 live citations captured so far — and we leave that number on the page. Sample or placeholder passages never appear here; when Perplexity, Gemini, or a manually logged ChatGPT/Claude citation names us (or a client domain), it gets a time stamp and a row.
Perplexity and Gemini run every Monday at 8am CT. ChatGPT and Claude are captured manually until their APIs expose stable citation metadata.
Methodology · running in the open
How we capture and verify every citation.
- 01
Prompts defined upfront
Every CTM client agrees on 20–30 target prompts at kickoff, the actual questions their customers ask AI engines. Recorded in the engagement doc and added to the tracker.
- 02
Cron runs every Monday at 8am CT
A scheduled job at
/api/cron/citation-checkqueries Perplexity and Gemini for every active prompt. ChatGPT and Claude are captured manually until their APIs expose stable citation metadata. - 03
Detection logic is public
A citation counts when the target domain or brand name appears in the AI response text OR when the target domain shows up in the engine's returned citation URLs. No fuzzy matching, no editorial discretion.
- 04
Everything logs to Supabase
Each run writes one row per (prompt × engine) to
citation_checks, engine, prompt id, response text, cited passage, citation URLs, run time. The wall reads live from that table. - 05
Citations are volatile, that's the point
AI engines re-train, prompts shift, competitors push back. Old citations sometimes disappear. The wall refreshes every 30 minutes so what's up here is current, not historical.
Want your business on this wall?
Start with a free 15 minute strategy call. We'll run live ChatGPT and Perplexity prompts on your industry during the call so you see exactly what we'd target first.
