Do Reviews, Backlinks and PR Affect Whether AI Recommends Your Business?

A business that asks 'why does AI recommend my competitor and not us' is usually really asking whether reputation — reviews, press, backlinks, general brand presence — plays a role in AI search the way it plays a role in human buying decisions and in traditional SEO. It is a fair question, and the honest answer sits between two unhelpful extremes: the claim that none of it matters because 'AI is different', and the claim that a review-buying or backlink campaign will reliably move AI recommendations.

Two related articles already cover why AI mentions one company over another and how AI systems decide what to recommend in general. This one focuses specifically on off-site signals — the things that happen away from your own website — and what can honestly be said about their effect, along with the real legal and policy risks of trying to manufacture them.

Start with how AI answers are actually grounded

FACT: Google documents that there are no special requirements to appear in AI Overviews or AI Mode beyond normal Search eligibility — a page must be indexed and eligible to appear as a normal result. Microsoft documents that Copilot grounds its web-based answers using Bing search results. Several AI assistants, including ChatGPT in search mode, retrieve from the live web via their own crawlers rather than answering purely from training data. The practical implication is that, where an AI answer is grounded in web search, it is very likely drawing on a pool of pages and sources that a traditional search engine has already crawled, indexed and to some extent ranked — which means the things that affect how search engines treat a page (including, historically, backlinks and reputation signals) are plausibly inherited by the AI layer on top, even though no provider states this as a rule.

Do customer reviews affect AI recommendations?

There is no published mechanism confirming this directly. The reasonable inference is indirect: reviews on platforms like Google, Trustpilot or industry-specific directories are themselves indexed, crawled content that search engines and some AI retrieval systems can read. A consistent pattern of genuine reviews describing what a business does and how well it does it gives both a search engine and an AI system another independent source confirming the same facts the business claims about itself — which is the corroboration effect described in the related article on why AI recommends one company over another. This is different from claiming that a star rating is a ranking input AI systems are known to use; it is not documented as one.

Do backlinks affect AI visibility?

Backlinks are a long-documented input to how Google ranks pages in ordinary search, which is the foundation several AI features sit on top of. There is no evidence that backlinks have a separate, additional effect specific to AI citation beyond whatever effect they already have on a page's underlying search ranking and discoverability. Treat backlinks as part of general SEO health rather than as a distinct 'AI visibility' lever, and be wary of any agency selling link-building specifically as an AI-visibility service — that framing is not supported by documented evidence.

Does PR and press coverage help?

Genuine press coverage does two things that are each independently plausible contributors: it usually creates an indexed page on a domain with its own search authority, and it adds another independent source describing the business in consistent terms. Both are consistent with how corroboration is understood to work for entity recognition (see the related article on how AI systems decide which companies to recommend). Press coverage is not a documented AI-ranking input in its own right, and a single piece of coverage is unlikely to change anything on its own — it is one data point among many a retrieval system might or might not use.

Does general brand authority matter?

'Brand authority' is not a term any AI provider defines or measures publicly. What can be said is that a business which is easy to identify unambiguously — consistent name, consistent description, present across multiple credible independent sources — is easier for any system, human or machine, to resolve with confidence. That is a reasonable restatement of the corroboration and entity-clarity points already covered elsewhere in this cluster, not a new or separate AI mechanism called 'brand authority'.

Myths vs reality

The risk in manufacturing these signals

FACT: Google's spam policies explicitly prohibit fake engagement and reviews, and participating in link schemes to manipulate ranking. FACT: in the UK, the Digital Markets, Competition and Consumers Act 2024 (DMCC) gives the Competition and Markets Authority stronger powers to act against fake and incentivised reviews, building on existing consumer protection law; businesses should treat fake or incentivised reviews as a genuine legal and reputational risk, not just a policy violation, and should check current CMA guidance before running any review-generation activity. Buying backlinks or running undisclosed paid placements as 'editorial' coverage carries equivalent risk under advertising standards and search-engine spam policies.

A competitor-comparison exercise worth running

1. 01Pick two or three competitors who are regularly mentioned in AI answers to your target buyer questions
2. 02List every independent source you can find describing each of them — reviews, directories, press, industry bodies, case studies on other people's sites
3. 03Compare consistency: do those sources describe the same service, location and specialism, or does the picture vary from source to source
4. 04Compare your own business against the same checklist, honestly
5. 05Prioritise closing the gaps that are genuinely within your control — claiming directory listings, responding to reviews, pursuing credible press opportunities — over anything that resembles manufacturing fake signals

This exercise will not tell you precisely why an AI system chose one business over another, because that mechanism is not published. It will usually show a real, fixable gap in independent corroboration, which is the most defensible lever available and the one this cluster consistently points back to.