Why the Old Definition of a 'Good Link' No Longer Works

For years, the SEO industry ran on a simple checklist: chase a high domain authority score, secure a dofollow link, and anchor it with an exact-match keyword. That formula no longer describes what a good link actually is.

According to Semrush's link building research, "the old criteria (a high domain authority score, a follow link, an exact-match anchor) have been replaced by something more contextual." The new signals are relevance, trust, context, co-citation proximity, persistence, and retrievability — qualities that matter to both traditional rankings and AI answer engines.

Nothing illustrates the shift better than a real example. Backlinko's analysis of high-quality backlinks examined a PR Newswire link carrying an Authority Score of 73 — a number most link builders would once have celebrated. The link failed on topical relevance, drove no meaningful traffic, and lacked genuine editorial placement.

In other words, a link can look excellent on a metrics dashboard while contributing almost nothing to how search engines and AI systems understand your business. Context now beats the href tag.

The contextual shift shows up clearly in large-scale correlation studies. An analysis of 1,000 domains across five AI platforms, cited in research on backlinks and AI citations, found that nofollow links correlated with AI visibility essentially identically to follow links — and image links correlated more strongly than text links.

That pattern suggests AI engines treat links less as PageRank equity and more as evidence that a brand exists somewhere reputable. The same research found branded anchors carried a 0.527 correlation with AI visibility, compared to just 0.218 for raw backlinks.

Based on the research, a link worth earning today does four jobs:

• Passes authority from a topically relevant, trusted source
• Sits inside genuine editorial content, not a syndicated press release
• Places your brand in context AI systems can learn from and retrieve
• Builds trust with human readers who may become customers

This is why practitioners like Collin Czarnecki of NOBLE Digital Studio describe successful link campaigns as "built like stories first and SEO assets second," noting in Backlinko's guide that accuracy and editorial relevance are becoming competitive advantages again.

For a local HVAC company or law firm, this redefinition is good news. A handful of relevant links can outperform thousands of generic ones. BrightLocal's local link building research confirms that sites with only a few links can outrank competitors with thousands when relevance is stronger — and links remain the fourth-biggest ranking factor for the Google Local Pack.

At AI SEO Consultants, this is exactly how link opportunities get evaluated: not by an authority score in a tool, but by whether the source is trusted, topically aligned, and placed in editorial context that both Google and AI answer engines can use as evidence. The old checklist measured what was easy to count. The new one measures what actually influences visibility.

Links Are the Threshold, Mentions Are the Decider

For years, the SEO industry debated which links move rankings most. The AI search era has reframed that question entirely: links get you considered, but mentions get you chosen.

The most important finding across recent large-scale studies is that links function as a threshold, not a dial. A baseline of referring domains earns your site a place in the candidate set AI systems draw from — but once you clear that bar, other signals determine whether your brand is actually recommended.

The data shows this clearly. According to SE Ranking's analysis, referring domains are the dominant predictor of ChatGPT citations and the second-strongest factor for Google AI Mode. But the returns are non-linear. Aggregated research across 75,000 brands found the biggest ChatGPT citation spike happens when sites cross roughly 32,000 referring domains (jumping from 2.9 to 5.6 average citations) — a clear "bend in the curve" rather than steady linear gains.

Kevin Indig summarizes it plainly: don't expect returns from linear growth, because a minimum investment is required before link building produces visible AI impact.

Two major studies seemed to disagree. Ahrefs found branded web mentions were the strongest AI visibility factor; SE Ranking crowned referring domains. The reconciliation is elegant: they measured different outcomes.

• Links predict being a source. Referring domains determine whether your URL appears in an AI's cited source list — the bibliography.
• Mentions predict being the answer. Branded mentions determine whether your brand name appears in the recommendation itself.
• The gap is massive. Branded web mentions correlate with AI visibility at 0.664, versus just 0.218 for raw backlinks — and branded anchors alone hit 0.527.

This distinction reshapes strategy. A brand with modest link authority but strong mention presence across reviews, community threads, and earned media can outperform a link-rich competitor in actual recommendations. The data backs this up: earned media accounts for 84% of AI citations, while having even 1–13 reviews lifts citation rates from 1% to 53.5%.

Build a diverse, topically relevant referring-domain base first — enough to clear the threshold. Then redirect investment toward the signals that decide recommendations: brand mentions, reviews, community presence on platforms like Reddit and Quora, and original data assets that publishers and models alike need to cite.

This is exactly how AI SEO Consultants structures visibility work for clients: establish the link foundation that gets a business into the candidate set, then build the mention and evidence layer — reviews, local citations, useful content — that makes AI systems confident enough to name the business in the answer. Neither half works alone.

One caveat worth stating: nearly all AI citation data is correlational, not causal, and the sources acknowledge this. Treat these patterns as directional guidance, not guarantees — but the convergence across multiple independent studies makes the threshold-versus-decider model hard to ignore.

The Link Types That Actually Move AI Visibility

The old playbook — chase high domain authority, secure dofollow links, optimize exact-match anchors — no longer predicts AI visibility. Research across 75,000 brands and 129,000 domains shows that links function as a threshold, not a dial: they earn a place in the candidate set of sources AI systems consider, but brand mentions and earned media decide which brand actually gets recommended.

Referring domains remain the dominant predictor for ChatGPT citations (SHAP weight 1.21) and the second strongest for Google AI Mode (0.56), yet the returns are non-linear — domains with fewer than 300 referring domains average ~2.5 AI Mode citations, while those above 24,000 average ~6.8. Once that threshold is cleared, the correlation shifts: branded web mentions correlate with AI visibility at 0.664 (Spearman) versus raw backlinks at just 0.218, and YouTube mentions lead all signals at ~0.737.

Link attributes matter far less than context. Nofollow links correlate with AI visibility essentially identically to follow links (0.340 vs. 0.334 Pearson), and image links correlate more strongly than text links (0.415 vs. 0.334). Branded anchors (0.527) dramatically outperform raw backlinks. This means unlinked brand mentions, podcast show notes, YouTube descriptions, and newsletter archives all count as retrievable evidence for answer engines.

• Editorially earned, topically relevant placements from trusted sources — local news, chambers of commerce, vertical directories
• Original data assets and tools on standalone URLs that become primary sources writers and LLMs must cite
• Earned media distribution — third-party publisher pickup delivers 239–325% citation lift over owned-domain publishing
• Review profiles and community presence — even 1–13 reviews lift median citation rate from 1% to 53.5%

Platform weighting differs: ChatGPT leans heavily on referring domains, while AI Overviews prioritize brand recognition, mentions, and semantic relevance. AI SEO Consultants helps clients build the diverse evidence layer — technical crawlability, editorial links, mentions, reviews, and fresh content — that AI systems discover and reuse across Google AI Overviews, ChatGPT, Gemini, Perplexity, and Copilot. The goal isn't chasing metrics; it's becoming the business AI recommends.

Technical and Local Link Fundamentals That Still Matter

While the AI search landscape keeps shifting, the technical plumbing underneath it hasn't changed at all — and local businesses that skip these fundamentals often undermine even the best link profiles.

Google's own documentation is unambiguous: search crawlers can generally only follow standard <a href> HTML links, so links rendered purely through JavaScript tricks or non-standard elements may never be discovered. According to Google Search Central, descriptive anchor text matters, and every page you care about should have at least one internal link pointing to it from another page on your site.

These aren't optional extras. Internal links distribute authority across your site and help both Google and AI crawlers understand which pages matter most. Linking out to trusted sources also builds your own credibility — a signal AI systems reuse when assembling evidence about your business.

For local rankings, links carry serious weight. BrightLocal's research identifies link signals as the 4th biggest ranking factor for the Google Local Pack — and here's the part that surprises most owners: sites with a handful of links can outrank competitors with thousands when relevance is stronger.

That finding has real implications for trades and professional firms. A plumber with links from the local chamber of commerce, a regional news story, and a sponsorship of a youth sports league will often outperform a competitor holding dozens of generic high-DR directory placements. Local relevance beats raw authority metrics.

The most valuable local link sources include:

• Chambers of commerce and local business associations
• Local news coverage and community event sponsorships
• Vertical directories specific to your trade or profession (e.g., FindLaw for legal)
• Local meetup groups and partnerships — relationships that yield what practitioners call "killer local links"
• Suppliers, manufacturers, and referral partners who list their dealers or affiliates

This quality-over-quantity logic extends into AI search. Research shows sites with fewer than 300 referring domains average roughly 2.5 AI Mode citations, so clearing a baseline threshold matters — but after that, context beats the href tag. Nofollow links correlate with AI visibility essentially identically to follow links, and branded anchors outperform raw backlink counts, per large-scale AI citation analysis.

Freshness is another technical fundamental with measurable payoff. Pages updated within the last three months earn 6.0 citations on average versus 3.6 for outdated pages — a reminder that link equity flows best to content that stays current.

In practice, this is why AI SEO Consultants treats technical accessibility, internal linking, and locally relevant citations as the evidence layer AI systems discover and reuse. Strong local links and crawlable structure don't just help you rank in the Local Pack — they form the foundation that determines whether your business becomes a candidate for AI-generated answers at all.

How to Prioritize Your Link Strategy (and What to Do After the Threshold)

You've built a handful of good links. Now what? The data says the answer changes once you cross a certain line — and most businesses keep spending in the wrong place.

Research across 129,000 domains shows link returns are non-linear: as Kevin Indig puts it, you need a minimum investment before seeing any real impact. Links function as a threshold, not a dial — they earn you a place in the candidate set of sources AI systems consider. What determines whether you're actually recommended is everything else: mentions, reviews, and community presence.

Before crossing that threshold, evaluate every link opportunity against four signals. Skip authority-score chasing; a PR Newswire link with Authority Score 73 can still fail on relevance, traffic, and editorial placement, according to Backlinko's analysis.

• Topical relevance — does the linking site cover your subject or serve your area?
• Source credibility — is this a publication, directory, or organization AI systems would trust?
• Editorial placement — is your brand discussed in context, not dropped in a listicle?
• Link status — remember, nofollow links correlate with AI visibility essentially identically to follow links, per Semrush's cross-platform study.

Once your referring-domain base is diverse and relevant, shift budget to what actually moves recommendations. Reviews are the highest-leverage move available: a business with no review profile sees a 1% median AI citation rate, while having just 1–13 reviews lifts that to 53.5% (https://www.geoptimizer.io/blog/do-backlinks-matter-ai-citations). No link campaign delivers that kind of jump per dollar.

Then invest in community presence and freshness. Heavy Quora and Reddit participation correlates with roughly 50–75% more citations — SE Ranking notes smaller sites can build authority there similar to what larger domains achieve through backlinks. And pages updated within three months earn 6.0 citations versus 3.6 for stale ones.

This is the managed approach we take at AI SEO Consultants: build the full evidence layer — crawlable links, reviews, citations, community mentions, fresh content — that AI systems discover and reuse, rather than chasing any single tactic. For local businesses especially, where links rank as the fourth biggest Local Pack factor, a handful of relevant local links beats thousands of irrelevant ones.

Not sure where you stand? Our free AI SEO Visibility Report shows how your business currently appears across Google, Maps, and the major answer engines — directional results, no black-box promises. It's the fastest way to see which side of the threshold you're on.