Do Backlinks Help You Show Up in ChatGPT and Other AI?

They probably help, but indirectly and without a guarantee. Public studies show a correlation between links, brand mentions and AI visibility, none proves that a link causes a citation, and AI companies do not reveal how they choose sources. Backlinks remain useful for discovery and ranking, as the guide on what backlinks are explains.

Quick summary

• Documented: Google uses links to discover pages and as a relevance signal. To appear in AI Overviews, the page must be indexed and eligible for a snippet. For ChatGPT search, OpenAI recommends allowing OAI-SearchBot.
• Correlation: brands mentioned more across the web appear more in AI Overviews, and sites with more referring domains tend to receive more citations in ChatGPT. Correlation is not causation.
• Hypothesis: that adding backlinks to a page raises the chance of an AI citation. We found no public causal test.
• Unknown: the weight of links in each AI, whether nofollow or sponsored count and whether the anchor influences retrieval.
• Recommendation: secure the technical foundation, invest in being cited where your audience is and treat links as part of SEO, not as a shortcut to AI citation.

How to read this guide

If you are a beginner, read the section on what is known, the evidence table and the recommendations, and skip the experiment design. If you already work in SEO, go to the studies, the limitations and the test. Terms such as grounding and query fan-out are in the glossary.

What is known, and not known, about backlinks and AI?

What is known is how Google and OpenAI describe access to content, and the correlations measured by third parties. What is not known is how each assistant weighs links when choosing sources. The table separates what has a basis from what is a guess.

Claim	Status	Basis
Google uses links to discover pages and as a relevance signal	Documented	Google's link best practices
To appear in AI Overviews and AI Mode, a page only needs to be indexed and eligible for a snippet, with no special optimization	Documented	Google's documentation on AI features
To appear in ChatGPT search, OAI-SearchBot must be able to access the site	Documented	OpenAI's documentation on bots
Brands mentioned more across the web appear more in AI Overviews	Correlation	Ahrefs study, 2025
Sites with more referring domains receive more citations in ChatGPT	Correlation	SE Ranking study, 2025
Ranking on Google leads to being cited by ChatGPT	Weak evidence	Low overlap in the Ahrefs study
Adding backlinks to a page raises the chance of an AI citation	Hypothesis	No known public causal test
nofollow or sponsored links count for AI	Unknown	No public data
The link's anchor influences AI retrieval	Unknown	Only a correlation with brand anchors

In plain English: imagine you want to know whether having many famous friends makes a restaurant get recommended by a critic. The data show that much-talked-about restaurants are recommended more. But that does not prove the famous friends cause the recommendation: the restaurant may be good, and that is why it is talked about and recommended.

What do the studies say?

Three studies from 2025 help calibrate the conversation. All measure association, not cause, and none covers your site in particular.

Ahrefs: 75,000 brands and AI Overviews

In the Ahrefs study on AI Overviews visibility, published on May 26, 2025, the Spearman correlation with brand presence in Google's AI Overviews was 0.664 for brand mentions across the web, 0.527 for brand anchors, 0.392 for brand search volume, 0.295 for referring domains and 0.218 for total backlinks. The authors themselves remind us that correlation is not causation and that all factors had a correlation from moderate to very weak. About 26% of brands had no mentions in AI Overviews at all.

The prudent reading: being talked about across the web goes along with visibility more than the raw count of links does. That is about Google, not ChatGPT.

SE Ranking: 129,000 domains and ChatGPT

In the SE Ranking study, published on November 24, 2025, 129,000 domains and 216,524 pages across 20 niches were analyzed, with a statistical model (XGBoost with SHAP) applied to the citations received in ChatGPT. Sites with up to 2,500 referring domains had, on average, 1.6 to 1.8 citations, and those with more than 350,000 had 8.4. Near 32,000 referring domains, the average almost doubled, from 2.9 to 5.6.

Two limitations matter. The first is that the lower band goes up to 2,500 domains: a small site, with dozens or hundreds, falls entirely in the first step, and the study does not say what happens when it gains 5 links. The second is that huge sites probably have many links because they are large, well known and heavily visited, and those factors move together, which prevents isolating the effect of the link. The authors warn that the factors are interdependent.

Ahrefs: 15,000 queries and overlap with Google

In the Ahrefs overlap analysis, from August 11, 2025, with 15,000 long-tail queries, the URLs cited by assistants overlapped little with Google's top 10 for the same query: 8.0% in ChatGPT (citations in the text), 6.1% in ChatGPT references, 8.6% in Gemini, 8.2% in Copilot and 28.6% in Perplexity. On average, about 80% of citations did not rank anywhere in the top 100. In Google's AI Overviews, 76% of citations came from the top 10.

Assistants break the question into several searches, which query fan-out explains, and what they read may not be what ranks for the original question.

For SEO professionals: the Ahrefs result weakens the chain "more links → better Google ranking → ChatGPT citation". If most ChatGPT citations do not rank in the top 100 for the original prompt, Google ranking is a poor proxy for visibility in it. The conclusion is not that links do not matter, but that the effect on ChatGPT, if it exists, goes through other paths. That is why it is better to measure AI visibility directly instead of inferring it from ranking. The guide on how LLMs search, RAG and grounding details how it works.

Why could links help? Four hypotheses

There are plausible mechanisms, but none is proven for AI assistants. Treating them as hypotheses avoids investing without criteria.

1. Discovery. Pages with no inbound links are harder to find. For Google this is documented (links lead the crawler to new pages). For other assistants, it depends on the indexes they use, which the company does not detail.
2. Ranking in search indexes. If an assistant queries a search index, backlinks that help rank in that index could help the page enter the candidate set. The weight of each index in each answer is not public.
3. Reputation and entity. Being cited by trustworthy sites, with or without a link, may help the system recognize the brand as a known source. It is consistent with the mention correlation, but it is only correlation.
4. Training data. Frequently cited pages are more likely to appear in training data. It is the weakest hypothesis for answers with search, because in those the sources seem to come from queries made on the spot, not only from training.

In plain English: think of links as roads leading to your shop. Roads help people get there, but no one knows whether the travel guide that recommends shops looks at the number of roads. Secure the road (links), the storefront (content) and the address on the map (indexing).

What should you do in practice?

Take care of the basics first, then invest in being cited where your audience is, and only then think about link volume. The order matters more than the quantity.

1. Secure the technical foundation. The page must be indexed and accessible. Allow OAI-SearchBot in robots.txt if you want to appear in ChatGPT search. The guide to robots.txt and AI crawlers explains each bot.
2. Write citable answers. A direct answer at the start, data with sources and a clear structure. The step-by-step is in how to show up in ChatGPT.
3. Pursue mentions, not just links. Press, partners and communities where people search. See unlinked brand mentions and Reddit, YouTube and Quora in AI citations.
4. Keep traditional link building as part of SEO, with relevant links and a natural profile. That is what is known to help on Google.
5. Measure AI visibility directly, with fixed questions, and not through ranking.

On buying links: Google considers links bought to manipulate rankings a violation of its spam policies, and asks for rel="sponsored" on paid links. Do not treat a paid backlink as a shortcut to appear in ChatGPT, because the effect is not proven. If you decide to buy for your SEO, Contentor's Brazilian backlinks are permanent, contextual links on Portuguese-language sites with DR 20+, from $49 per link in a white niche, with free replacement for 6 months. The decision and the risk are yours.

How do you test whether links help in your case?

With a simple experiment, with a control group and fixed, repeated questions. It does not prove cause, but it is much better than the impression that "after the link, the AI started citing us".

1. Choose comparable pages. For example, 10 pages on the same topic and at the same quality level, split into two groups of 5.
2. Define 20 questions per group, of comparable difficulty, that your customer would ask and that these pages would answer.
3. Measure the baseline. Ask ChatGPT each question at least 3 times (answers vary) and record the date, the question and whether your brand or URL appears.
4. Change only one thing. One group receives relevant mentions or links; the other, nothing.
5. Wait and repeat the measurement. A window of 8 to 12 weeks is a suggestion, not a rule.
6. Compare the groups and decide the criterion before seeing the result.

For SEO professionals: record each round as a binary event (the brand appeared or not) per question, along with the answer mode (with or without search), and compare the groups with a simple test of proportions. With 5 pages per group, statistical power is low, so treat the result as direction, not proof.

Cautions: with few pages, any result is an anecdote. Changes in the assistants' model and sources during the test also alter the numbers, so record the dates. If ChatGPT is the focus, Contentor has AI Visibility, which tracks whether ChatGPT cites your brand for the questions you choose (ChatGPT only). Other tools are in monitoring AI citations.

Hypothetical example: a scheduling SaaS for clinics

The data below is made up to show the reasoning. "ApptEase" has 10 content pages and wants to know whether more mentions help it get cited for questions like "what is the best scheduling software for a small clinic?".

Group	Intervention	Measure
A (5 pages)	Two articles in industry outlets and one participation in a specialized community	In how many of the group's 60 rounds (20 questions × 3) the brand appears
B (5 pages)	None	Same measure

Decision rule, defined beforehand: "if group A appears in at least 10 more rounds than B after 12 weeks, we repeat the experiment with other pages". If the difference is small, the result is inconclusive. If it is large, it is an indication and deserves replication, not a law.

Frequently asked questions

Do backlinks make ChatGPT cite me?

There is no proof of that. What exists is correlation: sites with more referring domains receive more citations in the SE Ranking study. But those sites are also larger and better known. Backlinks may help indirectly, through discovery and ranking, and nothing guarantees a citation.

Does a nofollow backlink help you appear in AI?

It is not known. Google treats nofollow, sponsored and ugc as hints for its own ranking, but AI companies do not say how they treat these attributes. A labeled link can still generate visits and mentions, which have value by themselves. See dofollow, nofollow, sponsored and UGC.

Are unlinked brand mentions worth as much as backlinks?

For AI visibility, the available data suggest that mentions go along with visibility more than backlinks do, but it is correlation and was measured in AI Overviews. For Google, the role of the link is documented (discovery and relevance signal), and the unlinked mention's is not in the same way. The ideal is to pursue both.

Does buying backlinks help you appear in ChatGPT?

There is no evidence that it helps. It may help traditional SEO, with Google's policy risk. For AI, focus on citable content, OAI-SearchBot access and real mentions. If you buy, do it for the right reason: speeding up pages that deserve to rank.

Do I need backlinks to appear in Google AI Overviews?

According to Google, there are no additional requirements or special optimizations beyond the page being indexed and eligible for a snippet. Links help with discovery and ranking, and that counts indirectly. Google's document does not make presence in AI Overviews conditional on having backlinks.

Next step

If your goal is to get cited, start with the most direct path: how to show up in ChatGPT. If you are weighing buying links for SEO in general, read the guide to buying quality backlinks before deciding.

Read also:

• Unlinked brand mentions
• Reddit, YouTube and Quora in AI citations
• How to measure AI visibility