Brand mentions vs backlinks in AI search

The “mentions versus links” debate has been quietly shifting for years. AI search has tipped it. In 2026, unlinked brand mentions in third-party content carry more weight than they did for classic SEO, and tech marketing teams are catching up to this slowly. The honest picture is that both still matter, the balance has changed and most B2B tech sites are under-investing in mentions.

This piece walks through what we believe holds up about each signal in AI search, where the evidence is thin and what we’d actually recommend doing.

Why this question matters more now

For traditional SEO, links were the dominant authority signal. Mentions without links helped a little, mostly through indirect means: brand searches, traffic and the eventual link that often followed. The model behind AI search is different.

LLMs are not running PageRank-style link graph computations on the fly. They are trained on, and ground answers in, a corpus that includes web content, third-party publications and structured knowledge. When the model decides which source to cite, it leans on a wider set of signals: who’s talking about your brand, what they’re saying, where the discussion appears and whether the entity is unambiguously identified. Links are still part of that picture. Mentions, particularly in trusted publications, now pull more direct weight.

This is why a competitor who’s been mentioned in TechCrunch, The Register, an industry analyst note and three named podcasts can outrank you in citation share even if their backlink profile is thinner. We’ve watched it happen on real client accounts.

What backlinks still do

Backlinks have not gone away. They still:

• Drive crawlability and discovery. New pages get found through links. LLM crawlers follow them like classic search bots do.
• Function as a structural authority signal. A site with strong, relevant backlinks tends to be a site whose pages get retrieved more often.
• Help with disambiguation. Inbound links from authoritative sources confirm which entity your domain represents.
• Drive direct human traffic. Which still converts.

The link-building tactics that worked for classic SEO still work for AI search to a meaningful degree. The link from a trade publication or industry directory has not lost its value. We covered the broader SEO foundations in our technical SEO audit checklist for tech sites.

What unlinked mentions now do

Unlinked mentions, particularly in named, credible publications, contribute to AI search visibility in ways they did not contribute to classic SEO. The mechanism is roughly:

• Entity reinforcement. The model builds a picture of who you are and what you’re known for from the wider corpus, links or no links.
• Topical association. A mention of “Acme Cloud Services, the Manchester MSP that specialises in legal IT” gives the model exactly the kind of attribute-rich association it needs to surface you for legal-IT queries.
• Trust transfer. A mention in a respected outlet contributes to how the model judges your domain, even without a link.

This is why analyst recognition, podcast appearances, trade-press coverage and inclusion in ranked lists (“Top 50 UK MSPs of 2026”) now have direct AI search value. Some of those will link, some will not. Both contribute.

What we’d argue from the evidence

A few claims we’d defend, with the appropriate hedging:

• Both signals matter, with different decay curves. A new link tends to take longer to influence AI citation than a new high-quality mention does.
• Mention quality matters more than mention volume. A single named mention in a credible publication outweighs ten low-quality directory listings.
• Linked mentions are still the strongest version. A mention in The Register that includes a link beats one that does not.
• Brand mentions on social platforms count less than mentions on independent publications. The signal is much noisier. Reddit is the exception worth tracking, as we cover in Reddit in AI search.

What we cannot prove cleanly is the exact weighting any individual LLM applies. The major models do not publish their grounding logic, and they change the weights without notice. Treat any precise claim about the ratio with scepticism.

Practical implications for B2B tech marketing

If both signals matter, where should a constrained tech marketing team spend its time?

We’d suggest a rough split that looks something like sixty per cent on mentions and forty per cent on traditional link-earning, with the caveat that the right split depends on where you’re starting. A site with no backlink foundation should spend more on links. A site with a solid backlink profile but a thin third-party narrative should spend more on mentions.

Specific tactics that have paid back for our clients:

• Trade press relationships. A working relationship with one or two journalists who cover your sector beats fifty cold pitches. The journalists who matter for B2B tech are a small group.
• Contributed articles in named publications. With your byline, your firm’s name in the bio and ideally a sentence in the body about who you are. Many of these will not link, and they still help.
• Analyst and ranking-list inclusion. Forrester, IDC, Channel Futures, MSP500 and the various UK-specific rankings. These are referenced heavily by LLMs.
• Named podcast appearances. With clean transcripts, episode descriptions and entity-rich introductions. We’ve watched a single podcast appearance materially shift citation share for a client.
• Genuine thought leadership. With a real viewpoint, by named people, on topics where you have credibility. We covered this from a different angle in content strategy for B2B tech.
• Comparison and review-site coverage. G2, Capterra, TrustPilot for product firms. Trade-specific directories for service firms. We’ve gone deeper on this in G2 and Capterra in AI search.

Each of these contributes to both signals at once. The trade-press article often gets a link. The podcast often does not. Both raise the floor.

How this fits with the rest of an AI SEO programme

Brand mentions sit alongside the other foundations we’ve covered:

• Solid technical foundations and content structure (technical SEO audit checklist for tech sites).
• Cite-friendly writing (writing content that AI search engines actually cite).
• Schema and entity disambiguation (structured data for AI search).
• Honest measurement (tracking AI search traffic and auditing your visibility in Copilot and ChatGPT).

Without those, mentions and links alone will not move the needle. With them, the third-party signal is often the bit that decides whether you become a default cited source or not.

How to track mention quality

A pragmatic approach we use with clients:

1. List your priority entity associations. What you want to be known for. “Manchester MSP for law firms.” “Mid-market SAP S/4HANA consultancy.” “Azure migration specialist for healthcare.”
2. Audit which entities currently appear alongside your brand in third-party content. Ahrefs, Semrush and Google Alerts all help. Manual review of the top fifteen mentions matters.
3. Score mentions for quality. Publication credibility, attribute-richness of the mention, whether it links.
4. Track quarterly. Both volume and quality. Aim for the quality curve to move up, not just the volume curve.

This is the kind of work that earns its place over a twelve to eighteen-month horizon. It does not produce dashboards that look impressive at the end of week one. It produces durable AI search visibility that is hard for a competitor to dislodge.

If you’d like a second opinion on your AI search strategy, drop us a line. The mention-versus-link rebalancing is one of the more interesting conversations we’re having with AI SEO clients in 2026, and the right answer is rarely the same twice.