Category playbooks
Why don't AI answers cite our Korean brand's US press coverage?
A US beauty article about a Korean brand can be live, accurate and flattering, and the answer an American shopper gets from ChatGPT can still leave the brand out. Getting the article published, getting it cited and getting the brand recommended are three separate events, and a press placement only proves the first one.
This playbook finds out which of the other two failed for each article, and who can fix it.
Sources checked on 6 October 2026.
Why can US press coverage be live and still missing from AI answers?
An AI answer cites the pages its engine retrieved for that one question, and a published article is not automatically among them. Both Google and OpenAI document conditions an article has to meet before it can be cited at all:
- Google AI Overviews and AI Mode: a page “must be indexed and eligible to be shown in Google Search with a snippet” to appear as a supporting link, and Google lists no additional technical requirements (Google, AI features and your website, updated 10 December 2025).
- ChatGPT search: “Sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers,” according to OpenAI’s crawler documentation, where the OAI-SearchBot setting is made in the site’s robots.txt (OpenAI, Overview of OpenAI crawlers).
Meeting those conditions makes an article eligible. It does not make the engine choose it. An article can also be read and still be skipped because it never names the exact product or never states the claim the shopper asked about.
What does a Korean brand need before checking its coverage?
Four records, gathered before anyone runs a question:
- A coverage inventory: every US article, with its URL, publication date, the product it names and how it attributes the brand.
- The buyer questions the coverage should support, written the way a US shopper asks them, not as keywords.
- An approved US product record: the product names, identifiers and the claims the brand’s compliance owner has signed off.
- A named contact at the PR agency who can take requests to publishers.
How do you find out why an article is not cited?
Seven steps, in this order. The first three measure, the fourth diagnoses, and the last three act.
- Run the buyer questions and keep the sources. Ask each question on ChatGPT, Perplexity, Claude, Gemini and Google AI Overviews, and record the wording, the engine, the date and every cited URL. Repeat the run over several days, because one run cannot tell a real absence from variation that reverses by itself.
- Sort the questions by whether coverage can move them. If an engine shows no sources under a question run after run, there is no cited page for an article to become. If the same domains are cited on most runs, those domains are the reading list to get onto. If the cited set changes on every run, collect more runs before choosing.
- Map each article to the questions. For every article, note which question it could answer and whether it appears among that question’s cited sources. Keep three cases apart: not cited at all, cited but credited to someone else, and cited for something that does not answer the question.
- Check each article on the five points in the table below.
- Fix what the brand controls first. The brand’s English product pages can link to the coverage, state the facts the article supports and use the same product names the article uses.
- Send specific requests for what the brand does not control. Each request names the question, the article, the fact to check, the change requested and the page owner. The PR agency takes it to the publisher, and the publisher decides.
- Run the same questions again. Keep the original answers and citations next to the new ones, and report each outcome on its own line.
What should each US article be checked for?
Five points. A failure on any one of them can keep an otherwise strong article out of the answer.
| Check | What to look for | Who can fix it |
|---|---|---|
| Product | The exact US product name, not only the brand name | The publisher, through the PR agency |
| Claim | The claim the buyer question asks about, stated in the article’s text | The publisher, through the PR agency |
| Date | A visible publication or update date, so an engine can tell the article from older sources | The publisher, through the PR agency |
| Readable text | The article returns as text to a crawler, appears in Google Search with a snippet, and the publisher’s robots.txt does not block OAI-SearchBot | The publisher, through the PR agency |
| Attribution | The product credited to the brand under the name US shoppers use | The publisher for the article, the brand for its own pages |
Four of the five rows belong to the publisher. That is why the brand’s own pages come first in step 5: they are the only part of the chain the brand can change without waiting for anyone.
What can a Korean brand fix without the publisher?
Its own English product pages. A page that links to the coverage, states the same facts in text and uses the same product name gives an engine a brand-owned source that agrees with the article. SearchD scopes this on-site work in one to two weeks, because it waits on nobody but the brand’s approval (how SearchD works).
The brand cannot edit a publisher’s article, its robots.txt or its date. Those go through the PR agency as requests, and work that waits on another party is scoped in months, not weeks.
How should the result be reported?
As three separate outcomes, never as one. A new link to the coverage, a corrected article and an AI answer that names the brand are different results, and each one gets its own line in the report.
The first re-measurement belongs 30 to 60 days after the on-site changes ship, because an engine has to re-crawl and re-answer before any change can show. Publisher-dependent work is scoped at three to six months (how SearchD works). Both are planning windows, not promised dates.
What mistakes make US coverage harder to get cited?
- Commissioning another article first. If the existing coverage cannot be read or matched to the product, a new placement can carry the same problem.
- Counting placements as proof. A list of placements says nothing about whether any engine used them.
- Treating Korean popularity as US evidence. Korean review counts and Korean press are context, not clinical proof or US sales evidence.
- Reading one before-and-after as the cause. Answers move on their own, and without a measured noise floor an uncontrolled change does not show that an edit worked.
Who can run this check for a Korean brand?
SearchD, an answer engine optimization agency for Korean brands selling in the US, runs this mapping as part of brand evidence and AI citations and hands each publisher request to the brand’s PR team. The free report shows which sources answer a brand’s buyer questions today.
Related questions
Does checking why coverage is not cited replace our PR agency?
No. The PR agency keeps its role and its relationships with publishers, and it is the one that takes correction requests to them. New pitching, paid placement and content rights are separate decisions. Whether a publisher accepts a request, and whether an engine then cites the article, stays outside the control of the brand, the agency and SearchD alike.
How long before AI answers reflect a fix?
Changes on the brand's own pages can ship in one to two weeks. An engine has to re-crawl and re-answer before any change can show, so SearchD scopes the first re-measurement at 30 to 60 days, and work that waits on publishers at three to six months. These are planning windows, not promised dates for a result.
Is a press placement proof that AI assistants use it?
No. A placement shows that the article was published. Whether an engine cited it for a buyer question, and whether the answer then named the brand, are separate outcomes that only a recorded run of that question can show.
Can SearchD run this check for a Korean brand?
Yes. SearchD, an answer engine optimization agency for Korean brands selling in the US, compares a brand's coverage inventory with the sources cited for the buyer questions agreed with the brand, checks each article, proposes edits to pages the brand controls and writes owner-specific requests for the PR team to take to publishers.