How do I run outreach to get added to pages AI cites?

Lock a set of prompts your buyers ask and run them on ChatGPT, Perplexity, Gemini and other engines, logging every cited URL. Rank those pages by citation frequency, competitor presence and feasibility, then send each publisher one specific, checkable correction or addition with the exact wording and proof. Disclose any offer. After the change goes live, re-run the same prompts and compare the answers with your baseline.

Can you actually influence what AI search engines say about a brand?

Partly, and only indirectly. You cannot edit an engine's answer, but you can work on the third-party pages it retrieves and cites. Ahrefs reported in May 2025 that web mentions correlate more strongly with AI Overview brand visibility than backlinks do. That is a correlation, not a guarantee: a corrected or added mention can change an answer, and no published study gives an expected success rate.

Do I have to disclose a paid placement in a listicle that AI engines cite?

Yes, plan for it. The FTC's Endorsement Guides require connections between advertisers and endorsers to be disclosed clearly and conspicuously, and the UK Advertising Standards Authority applies comparable rules. Separately, Google's spam policies treat paid links that pass ranking credit as link spam, so a paid link should carry the sponsored link attribute. Ask the publisher to label the placement on the page.

Can I ask Wikipedia to add my brand to an article AI engines cite?

You can ask, but avoid editing the article yourself. Wikipedia's conflict-of-interest guideline asks editors with a conflict of interest to propose changes on the article's talk page through an edit request, so that other editors can review them. Support the request with independent, reliable sources, disclose your connection, and accept that volunteers decide whether the change is made.

How to Run Outreach to Get Your Brand Into Pages AI Engines Cite

Q: How long does it take for a new mention to show up in ChatGPT or Perplexity answers?

No vendor publishes a delay, and it can differ by engine and by page. Record the date the mention goes live, then re-run the same prompts on each engine and log the date the updated page is first cited and the date your brand is first named. Make sure the page does not block the engine's crawler, because OpenAI states that sites opted out of OAI-SearchBot are not shown in ChatGPT search answers.

To get your brand added to the pages AI engines cite, find the URLs that ChatGPT, Perplexity and Gemini cite for your target prompts, rank them by competitor presence and feasibility, and send each publisher one specific correction or addition, disclosing any offer. Then re-run the same prompts to check whether the answer changed.

What AI citation outreach is, and how it differs from link building

AI citation outreach means getting your brand named, accurately, on third-party pages that AI engines already retrieve and cite. Success is a sentence on that page that names and describes your brand correctly. A link may come with it, but it is not the goal.

The case for third-party mentions rests on correlation data. In a study published on 26 May 2025, Ahrefs reported that "web mentions (0.664) correlate much more strongly than backlinks (0.218)" with brand visibility in Google's AI Overviews. That is a correlation across brands, not proof that one new mention moves an answer, but it points effort at mentions over links.

Link building and citation outreach overlap: a placement on a well-read roundup can also earn a backlink and its SEO value. They diverge on target selection. Link building picks pages for domain metrics. Citation outreach picks a page because an AI engine cites it for a prompt you care about, whatever its domain metrics.

The brand must appear in the cited text itself. Semrush's ghost citations study, published on 9 June 2026, found that "62% of AI citations are ghost citations. Your site gets a source link, but the AI never says your name in the answer." For the distinction in detail, see how an AI mention differs from an AI citation.

One limit up front: no published study reports reply or placement rates for this kind of outreach with a stated sample size and method, so this article gives no expected rate.

How to find and prioritise the pages AI engines already cite

The target list comes from the engines, not from a domain database, and the workflow below needs only a spreadsheet.

Collect cited URLs for a fixed prompt set

Write the prompts your buyers ask, such as category questions, comparisons and alternatives to a competitor, and lock the wording. Run each on ChatGPT, Perplexity, Gemini and any other engine your audience uses. For each answer, record the date, engine, prompt, every cited URL and every brand named. Run the set more than once; a URL cited only once may be noise.

Retrieval rules differ by engine. OpenAI's crawler documentation, verified on 13 October 2026, states: "Sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though can still appear as navigational links." A page that blocks that crawler is a weak ChatGPT target, however often another engine cites it.

When one competitor keeps appearing, trace a competitor's AI mention back to the page behind it and add that page to the list.

Score each page on competitor presence and feasibility

Score each cited URL on these criteria:

Citation frequency: how many prompts and engines cite it across your runs.
Competitor presence: competitors are named on the page and your brand is not. That gap is a strong reason to pitch.
Feasibility: a named author or editor, a contact route, a recent update date and a format that accepts additions.

Then filter by page type. Listicles, roundups and comparison pages are the most pitchable, because adding or correcting an entry fits how they are maintained. Opinion essays and news reports rarely accept additions.

As a hypothetical example, not a measured figure: working alone, collecting the citations might take an afternoon and scoring them another.

Adjust the target list for local, ecommerce and YMYL niches

Pitchable page types shift outside business-to-business software:

Local: cited pages are often directories, review platforms and city roundups. The ask is a complete, accurate listing, not an editorial addition.
Ecommerce: comparison and review pages are common. Publishers want product access, specifications and test data they can verify.
YMYL (health, finance, legal): only qualified expert sources are credible. Offer a named, credentialed expert or published data, and expect a slower review.

Which outreach ask fits which cited page

Each page type has an ask a publisher can accept without changing the page's purpose. Every ask rests on material the publisher can check: original data, public documentation, named expert input.

Page type	The ask	What you offer	Disclosure or policy risk
Page with outdated brand facts	Correction of one fact	The current fact and a public page confirming it	Low
Listicle or roundup	Addition to the existing list	A drafted entry, pricing, a demo account	Disclose any free access given in exchange
Comparison page	A new row or criterion	Verifiable specs, documentation, test data	Low if every fact is checkable
Report or explainer	Expert quote or data contribution	Named expert input or an original dataset	Disclose the expert's affiliation
Peer or partner blog	Content exchange	A guest article or co-written piece	Excessive link exchanges are link spam
Site with a sponsored programme	Paid placement	Payment	Clear disclosure and `rel="sponsored"` on the link
Wikipedia (unpitchable)	Edit request on the talk page	Independent, reliable sources	Conflict-of-interest guideline discourages direct editing
Paywalled outlet (unpitchable)	Correction through the editorial contact	A documented factual error	No route for promotional additions
Competitor-owned page (unpitchable)	None	None	Publish your own primary source instead

Some cited pages need a different route. Wikipedia's conflict-of-interest guideline, verified on 13 October 2026, asks editors with a conflict of interest to "propose changes on talk pages" through its edit request template, "so that they can be peer-reviewed." Cite independent, reliable sources. Paywalled outlets rarely take additions but often have a corrections route for factual errors. Competitor-owned pages are off the list: answer them by publishing a better primary source of your own.

When a cited page is wrong or negative about you, you have options beyond editing it. Send a correction request naming the error, the correct fact and the document that proves it. Publish a primary source an engine can retrieve, such as a dated changelog, and check crawlers can reach it. For a review or opinion piece, ask for a right of reply rather than a rewrite.

How to write the pitch and follow up within email rules

Open with one verifiable correction or addition

A publisher should be able to accept your pitch in a few minutes. Build it in this order:

The fact: one specific, checkable statement, such as an outdated price or a missing product on the list.
Why it matters: what the page's readers get wrong while the fact stays uncorrected.
The exact wording: a sentence or table row they can paste.
The proof: a link to public documentation, data or a named expert's profile.

Personalise it: name the page, quote the line to change and address the editor by name. Leave out company history and link requests.

Follow-up cadence and volume

A common practice is one follow-up about a week later, adding something new, then stopping. That is a courtesy convention, not a measured optimum. Keep volume low and check each message by hand: every pitch rests on a page-specific fact, and one unchecked fact costs the publisher's trust. Bulk sending also draws spam complaints that hurt your sending domain.

CAN-SPAM and GDPR rules for publisher outreach

A pitch that promotes your product can count as commercial email. The FTC's CAN-SPAM compliance guide, last edited in January 2024 and verified on 13 October 2026, requires accurate header information, honest subject lines, identification of the sender, a valid physical postal address and a working opt-out. It says you must "honor a recipient's opt-out request within 10 business days."

In France, where GDPR applies, the CNIL's page on email prospecting, dated 10 June 2026 and verified on 13 October 2026, allows business-to-business prospecting without prior consent when the message relates to the recipient's profession, provided the person is informed about the use of their data and can object. Put an opt-out in every message and say where you found the address.

Paid and exchanged placements: disclosure and spam-policy risk

An earned mention on a cited page can carry SEO and backlink value as well as AI visibility. That value is what spam policies regulate once money or an exchange enters the deal.

Google's spam policies for web search, verified on 13 October 2026, treat buying or selling links for ranking as link spam and list "excessive link exchanges ('Link to me and I'll link to you') or partner pages exclusively for the sake of cross-linking." A paid link is acceptable when qualified with rel="sponsored" (or rel="nofollow"), so it passes no ranking credit.

Disclosure is a separate duty. The FTC's Endorsement Guides cover connections between advertisers and those who promote them; the FTC's guidance page, verified on 13 October 2026, notes the revised guides include "a new definition of 'clearly and conspicuously'". If you pay for a slot, give free access for coverage or swap content, expect the publisher to disclose it clearly on the page. In the UK, the Advertising Standards Authority applies comparable rules.

Publishers may refuse a request whose only purpose is to change what a chatbot says. Frame every ask around the page's readers, and if the change would not improve the page for a human reader, do not ask for it.

How to tell whether a placement changed the AI answer

Once a placement is live, re-run the same locked prompt set on each engine with the same wording, and log the date, cited URLs and brands named. Recheck later that the mention is still live, because listicles get rewritten and entries vanish. Once this becomes routine, you can automate the re-checks of your AI visibility.

Answers vary between runs, so one before-and-after comparison proves little. Add control prompts whose cited pages you did not touch, and run them before and after alongside the target prompts. If your brand appears more often in target prompts while controls stay flat, the placement is a plausible cause. If both move, something else changed, such as a model update.

No vendor publishes how long a newly edited page takes to appear in ChatGPT, Perplexity or Gemini answers, so do not plan around a fixed delay. Log the go-live date, then, per engine, the date the updated page is first cited and the date your brand is first named.

Expect the metrics to diverge. A citation count can stay flat while a visibility score moves, because visibility depends on whether the answer names your brand, not only on which URLs it cites. An already-cited page that now names you can raise visibility and leave the citation count unchanged.

Where to start this week

Lock a couple of dozen prompts your buyers ask, with fixed wording, plus a few control prompts.
Run each prompt on ChatGPT, Perplexity, Gemini and your other engines, recording the date, cited URLs and brands named.
Score the cited URLs on citation frequency, competitor presence and feasibility.
Tag every target as pitchable, unpitchable or requires disclosure before writing any email.
Shortlist a handful of pitchable pages where competitors are named and you are not.
Prepare one correction-led pitch per page: the fact, why it matters, the exact wording, the proof.
Log send dates and save the baseline answers for target and control prompts before anything goes live.

To build your first target list, run a free Namedrop scan to see which pages AI engines cite for your prompts and which competitors they name.