Every year, someone declares link building dead, and every year agencies keep buying links. Both camps are partly right. The tactics that manufactured rankings are dying; the tactics that earn genuine third-party endorsement are becoming more valuable, because AI search now reads the web the way a sceptical editor does — checking not just whether a site links to you, but whether real publications talk about you, cite you, and trust you. Link building for 2027 is therefore less about volume and more about a simple test: would this link still be worth having if search engines did not exist?
This article gives you VonixCo’s own keep-and-bury list for 2027, built on the industry’s current conversation — including Search Engine Journal’s October 2026 “Link Building Graveyard” webinar, in which Resolve’s Michael Johnson set out which tactics he would fund heading into 2027 and which he would leave behind — plus a 90-day plan for resetting a programme around the keep list. Nothing here is a ranking guarantee. It is a practical framework for spending a link budget where it still earns something.
The Test Every 2027 Link Should Pass
Johnson’s central test, from the SEJ webinar, is worth adopting verbatim as a filter: a worthwhile link should benefit three parties — the brand, the publisher, and the publisher’s audience. If only the brand benefits, the placement is a favour, not an endorsement, and both search engines and AI systems are getting better at telling the difference. Run every prospective placement through that test before you pay for it or pitch it.
Two supporting shifts make the test more urgent in 2027. First, Semrush’s 2026 link building guide splits tactics into “link producers” — outreach, broken link building, resource pages — which are accessible and high-volume, and “association producers” — linkable assets, digital PR, becoming a cited source — which are harder but produce the brand association that search engines and AI systems use when deciding which brands to rank, mention and cite. A programme built only on link producers is easy to start and easy to copy. Second, Ahrefs’ 2026 mention studies found that brand mentions across the web track AI visibility far more closely than raw backlink counts do — branded mentions correlated 0.664 with AI Overview visibility against 0.218 for backlink count in their 75,000-brand study, and their later 260-million-prompt study found mention types correlating between 0.59 and 0.75. Correlations are not causation, and the studies cover AI answers, not traditional rankings — but the direction is consistent: being talked about matters alongside being linked to.
What to Bury in 2027
Private blog networks. Networks built solely to sell links leave footprints — shared hosting, thin content, identical link patterns — and they offer nothing to a publisher’s audience. The risk is not just a wasted fee; a visible pattern of network links can taint the whole profile. Bury them.
Link swaps at scale. A genuine reciprocal link between two sites that actually reference each other is normal. A scheme that trades links in bulk with unrelated sites is a different thing, and it fails the three-party test: the audience gains nothing. Bury it.
Forced commercial anchors. Forcing exact-match commercial anchor text into articles is how the anchor text profile that our anchor text strategy guide warns about gets built — unnatural, detectable, and increasingly pointless as AI systems weigh mentions and context over anchor text alone. Bury forced anchors; let editors choose natural wording.
DA/DR-only prospecting. Buying placements by a single third-party metric is the habit SEJ’s webinar explicitly called out as dying. A high-metric site with no real audience, no topical relevance and no editorial standards is a billboard in an empty field. Our vetting checklist covers the fuller test: relevance, real traffic, editorial standards, link profile and AI visibility.
Generic roundup-addition requests. “Please add us to your listicle” with no reason the list gets better is the lowest-effort outreach in the industry, and editors treat it that way. If you want a listicle placement, earn it with a genuinely useful resource — or write the comparison content yourself on a publisher that accepts contributions.
What to Keep in 2027
Expert guest contributions. A genuinely expert article, written for a publisher’s readers, with a contextual link where it helps the reader, remains on every serious keep list — including Johnson’s. The difference from the old guest-post mill is the standard: subject expertise, original perspective, and a piece that would make sense even without the link. See our complete guide to guest posting for how that standard works in practice.
Unlinked-mention conversion. Sites already mentioning your brand without linking are warm prospects: the editorial decision to mention you is already made. Our unlinked brand mentions guide walks through finding and converting them, and the Ahrefs data above gives them a second justification — the mention itself already contributes to AI visibility, and the link adds the traditional signal on top.
Expert commentary and digital PR. Journalists and editors need quotable experts. Original data, a clear opinion, or a fast, useful comment on a breaking story earns coverage that no outreach template can buy. This is the most effort-intensive keep item and the hardest to fake — which is exactly why it survives.
Linkable assets and resource pages. A genuinely useful asset — original research, a calculator, a definitive guide — earns links from resource pages and citations from AI answers alike. Semrush’s “association producers” category exists for this reason: assets create the association, not just the link.
Broken link building and lost-link reclaim. Both still work because both help the publisher: fixing a dead link or restoring a lost one improves their page. They are link-producer tactics, so they scale — but only when the replacement content is genuinely better than what was lost.
The 90-Day Reset Plan
Johnson’s webinar organised the reset into a sequence, and it adapts cleanly to an agency or in-house programme:
- Weeks 1–2: audit. Score your existing profile against the keep/bury lists. Flag network links, swap schemes and forced anchors; do not panic-clean everything at once, but stop renewing the bury list immediately.
- Weeks 3–4: baseline. Record where the brand currently appears in AI answers for its core buying prompts, not just where it ranks. That baseline is what the keep list is meant to move.
- Weeks 5–8: prospect and build. Identify publishers and resource pages that pass the three-party test, and create the asset those audiences actually need — one strong asset beats five thin pitches.
- Weeks 9–12: pitch and measure. Pitch expert contributions and mention conversions, track live placements (verified URLs, not tracker estimates), and re-run the AI-answer baseline. Judge the programme on placements earned and mentions gained, with ranking movement as a lagging indicator — our link building timeline sets that expectation honestly: live links from week three onward, ranking movement typically from month two or three.
Frequently Asked Questions
Is guest posting dead in 2027?
No — but bulk, low-quality guest posting is. Expert contributions on relevant publications with real audiences remain on every credible keep list. The distinction is editorial quality, not the format.
Should we stop tracking DA and DR entirely?
No. They are useful screening signals. The mistake is using them as the only criterion. Relevance, real traffic, editorial standards and audience fit decide whether a placement is worth having; a metric score alone does not.
How much of the budget goes to “association producers”?
There is no universal ratio, but a programme spending everything on easy link-producer tactics and nothing on assets, commentary or digital PR is buying the part of link building that AI search values least. A reasonable starting split for most agencies is roughly half the effort on producers, half on association — adjusted by what the brand already has.
Do mentions really replace links?
No. Mentions complement links. Ahrefs’ data shows mentions tracking AI visibility more closely than backlink counts, but that is a correlation across thousands of brands, not a reason to stop earning links. The 2027 programme earns both: mentions through commentary and contributions, links through the same placements.
VonixCo builds link programmes around the keep list — niche-relevant guest contributions on real sites, unlinked-mention conversion, and verified live placements with honest timelines. See our pricing guide or reach us at vonixco.com or on WhatsApp at +92 340 0096621.
Source note: keep/cut framework and 90-day sequence summarised from Search Engine Journal’s recap of “The Link Building Graveyard: What To Bury & What Still Wins” (searchenginejournal.com); link-producer/association-producer split from Semrush’s “Link building for SEO: What works in 2026” (semrush.com); mention correlations from Ahrefs’ 75,000-brand AI Overview study and 260-million-prompt mention-type study (ahrefs.com), reported as correlations, not causation.
