Pre-flight checks (day 1)

  1. robots.txt: OAI-SearchBot allowed; GPTBot decision documented; ChatGPT-User allowed.
  2. Bing Webmaster Tools: site verified, sitemap submitted, IndexNow enabled.
  3. Baseline run: 12-prompt brand test (identity/category/comparison/reputation/boundary), 3 runs each, sources panel screenshotted.

Source priorities

  1. Owned pages with answer blocks on the top-10 buyer questions.
  2. Review/comparison platforms — complete profiles; ChatGPT's synthesis leans on consensus sources for "best X" answers.
  3. Wikipedia/Wikidata for entity facts where eligible.
  4. Community presence (Reddit/niche forums) for experiential questions.

30-day plan

WeekMovesCheckpoint
1Pre-flight fixes; Bing indexation confirmedSite:query returns pages in Bing
2Answer-block reformat of top 10 pages; Organization schema deployedPages re-fetched; live answers begin shifting
3Directory/review profile completion; two corroborating placementsCited-URL inventory shows new sources
4Re-run baseline prompt set; log deltasVisibility movement documented per prompt

Failure modes specific to ChatGPT

  • Memory-vs-retrieval drift: answers from training memory lag your corrections. Symptom: Perplexity already right, ChatGPT still wrong. Fix: keep corroborating; memory catches up with model refreshes.
  • Entity collision: answered as your namesake. Fix: differentiation descriptors + Wikidata.
  • Sources-panel absence: mentioned but never cited. Fix: passage structure on owned pages — you're being remembered, not read.

Frequently asked questions

How long until ChatGPT reflects my fixes?
Live retrieval (search-enabled answers): days to weeks after Bing indexes the changes. Training-memory answers: one or more model refresh cycles. Track both answer types separately — they move on different clocks.
Does ChatGPT advertising affect answers?
Paid placements in ChatGPT are labeled ads, separate from organic synthesized answers. GEO work targets the organic answer layer; ads don't buy citations.