• Many AI recommendations cite third-party pages, not the brand's own site.
• Find the sources cited for your priority prompts and work to be included in them.
• Mentions without links still count, because models learn from text.
• Reviews, community discussion and independent comparisons carry particular weight.
• Authentic participation works. Spam and fake reviews backfire.

Ask an assistant to recommend a product and look at the sources. You will usually find review sites, "best of" lists, Reddit threads, YouTube videos and trade publications. The brands being recommended often appear because other people wrote about them. Your own website tells a model what you claim. The rest of the web tells it whether to believe you.

Why off-site matters so much

• Retrieval favours independent sources for evaluative questions such as "best", "top" and "X vs Y".
• Training data reflects the whole web. A brand discussed widely and consistently becomes strongly associated with its category.
• Consensus builds confidence. When several unconnected sources say the same thing, a model is more willing to repeat it.
• Links are not required. For a language model, the words themselves are the signal.

Step 1: find out who gets cited

• List the prompts that matter to your business, especially recommendation and comparison questions.
• Run them across the main assistants and record every cited source.
• Group the sources: review platforms, publishers, forums, video, competitors, reference sites.
• Note which mention you, which mention only competitors, and which appear again and again.

That list of recurring sources is your target list. It will differ by platform, industry and country.

Step 2: earn your place in those sources

Listicles and comparison articles

If a "best tools for X" article is cited often and leaves you out, contact the author with a clear reason to include you: a demo, data, a distinctive feature. Keep it honest. Undisclosed paid placement is risky and may breach advertising rules.

Review platforms

Maintain complete, accurate profiles on the platforms relevant to your sector, such as G2, Capterra, Trustpilot, Tripadvisor or Google reviews. Encourage real customers to leave detailed reviews, and respond to criticism professionally.

Communities

Reddit, Quora, specialist forums and LinkedIn discussions are heavily cited. Take part as yourself, declare your affiliation, and be useful. Communities are quick to spot and punish disguised promotion.

Digital PR

Original research, surveys, data studies and expert commentary earn coverage in publications that models trust. A story journalists want to write produces mentions, links and authority all at once.

Expert contributions

Guest articles, podcast appearances, conference talks and quotes in other people's content attach your name and expertise to your topic across many sources.

Video

YouTube is one of the most cited domains in AI answers. Reviews, tutorials and explainers on your own channel and on creators' channels extend your footprint.

Reference sources

Wikipedia and Wikidata are influential. Engage only within their rules: notability is required, and editing about your own organisation is a conflict of interest.

Step 3: shape what is said

The context of a mention matters as much as its existence. Aim for mentions that:

• Place you in the right category.
• Link you to the audiences and use cases you want.
• Include accurate facts about features, pricing and positioning.
• Sit alongside the competitors you want to be compared with.

Make this easy by publishing clear, quotable descriptions and fact sheets on your own site.

Sentiment

Models summarise the balance of opinion. A pattern of complaints about support or pricing will surface in answers about you. Fix the underlying problem where you can, respond publicly where appropriate, and make sure up-to-date positive evidence exists.

What not to do

• Fake reviews or undisclosed paid reviews, which are illegal in the UK and other markets.
• Mass-posting in forums through fake accounts.
• Flooding the web with self-published "best of" lists that rank yourself first on thin evidence.
• Buying mentions on link-farm sites.
• Publishing misleading claims about competitors.

Common mistakes

• Measuring PR only by links and missing unlinked mentions.
• Targeting prestigious publications that models rarely cite for your topics.
• Ignoring review platforms because they send little direct traffic.
• Leaving outdated information on third-party profiles.
• Treating this as a one-off campaign instead of ongoing work.

How to test

• Re-run your prompt set regularly and track your share of mentions against competitors.
• Monitor which third-party sources are cited and whether you now appear in them.
• Use brand monitoring tools to track mention volume and sentiment.
• Watch branded search volume, which often rises as AI exposure grows.

Add your title here

This is a paragraph. Writing in paragraphs lets visitors find what they are looking for quickly and easily. Make sure the title suits the content of this text.