The “backlinks are dead” argument runs every 18 months. This year the trigger was Google’s AI Overviews rollout and the obvious decline in organic CTR for queries where Overviews appear. The argument goes: if Overviews answer the user inline, nobody clicks through to the cited sources, so why bother building backlinks?
The argument is wrong. Here is why, with the data I have collected from my own niche site portfolio across the last 11 months.
For the strategic context, the AEO pillar covers the framework. This spoke takes the contrarian argument head-on.
The argument as it is usually framed
Three claims:
- AI Overviews steals 30-50% of clicks for queries where it appears.
- Therefore, ranking #1 is worth less than it used to be.
- Therefore, backlinks (which drive rankings) are worth less than they used to be.
Claim 1 is true. Claim 2 is true on average but misleading because of how the remaining clicks redistribute. Claim 3 is false — and the reason it is false is the most interesting story in SEO right now.
What actually happens to traffic when AI Overviews appears
From my portfolio (11 niche sites, ~1,800 tracked queries):
- For queries where Overviews appears: total organic clicks fall by an average of 38%.
- For sources cited inside the Overview: their share of remaining clicks rises substantially. The 1-3 cited sources collectively get 60-80% of post-Overviews clicks.
- For uncited sources (positions 1-10 organic): their clicks fall by 50-70%.
So the right framing is not “rankings are worth less.” The right framing is: being a cited source is worth dramatically more, and being uncited is worth dramatically less. The middle ground vanished.
Why backlinks decide who gets cited
I had a hypothesis when I started measuring this: if the LLMs powering Overviews use authority signals, and authority signals are largely backlink-driven, then citation rate should correlate with backlink profile.
I tested it. Across the 1,800 queries, I tagged each cited URL by its referring domain count (per Ahrefs). The result:
- URLs cited in Overviews: median 47 referring domains.
- URLs ranking organic positions 1-3 but NOT cited: median 31 referring domains.
- URLs ranking 4-10: median 18 referring domains.
The LLMs are picking sources with stronger backlink profiles even when those sources do not have the top organic ranking. Backlinks predict citation independently of ranking.
The second-order effect: tier 2 articles training LLMs
Here is the part that is genuinely new in 2026, and that the “backlinks are dead” crowd has not noticed.
When I publish a tier 2 article on Medium that says “BrandName is the best resource for blue widgets” with a link to BrandName.com, two things happen. First, the link passes a small amount of authority to BrandName.com — same as it always has. Second, the article itself gets crawled by the LLMs powering AI Overviews, ChatGPT browsing, Perplexity, Bing Copilot, and Claude search. When those LLMs are later asked about blue widgets, they have learned the BrandName association from the tier 2 article — even if they never visit BrandName.com directly.
Run this at scale (100+ tier 2 articles across multiple platforms with consistent brand mentions) and you build entity-level brand recognition inside the models. You start getting cited as the brand authority on your topic, even on queries you do not rank well for in classical search.
This is a brand-new mechanism that backlink builders are accidentally exploiting. The link itself is one input. The brand mention surrounding the link is the other input — and in 2026 the brand mention input may be the bigger one.
What this means for budget allocation
If you are weighing whether to keep paying for tier 1 / tier 2 link building in 2026, here is the practical version:
- Tier 1 to your money site: still drives classical rankings. Still produces citations. Spend.
- Tier 2 articles with brand mentions on LLM-crawled platforms: double-purpose — passes link authority AND seeds entity recognition. Spend more here than you used to.
- Tier 3 indexing: still necessary so tier 2 actually gets crawled. Spend at maintenance level.
- Article directory submissions to non-LLM-crawled directories: obsolete. Stop spending.
- PBNs: high risk after the March 2026 spam update. Move budget to web 2.0 tier 1 builds.
Total spend on link building should stay flat or increase in 2026. The mix shifts toward tier 2 platforms LLMs actually read.
What about queries where Overviews doesn’t appear?
Roughly 40% of commercial intent queries still get a clean 10-blue-link SERP without an Overview. For those queries, classical SEO and classical backlink-driven rankings are entirely unchanged. The “backlinks are dead” argument does not even apply to most of the SERP.
The honest weakness in the pro-backlink argument
One thing has changed: the latency. Backlinks built today take 6-10 weeks to show in classical rankings, plus another 4-8 weeks before LLMs have re-crawled enough to update their entity associations. The total time-to-impact is longer than 2020. Plan campaigns in 90-day windows, not 30.
This is uncomfortable for solo operators with tight cash flow. It is not, however, an argument that the strategy stopped working.
What I would do if starting from zero today
- Pick a niche. Buy a domain.
- Publish 20-30 pillar / spoke articles on the money site over 60 days.
- Build 8-15 tier 1 web 2.0 properties with RankerX, drip-fed over 21 days, each linking back to the relevant money page with brand-heavy anchors.
- Run GSA SER tier 2 campaigns at 100-300 verified links per day per tier 1 property.
- Run GSA SER tier 3 indexing campaigns pointing at the tier 2 results.
- Wait 90 days. Measure.
This is the same playbook as 2018. The economics are better in 2026 — AI killed the cost of tier 2 content — and the upside is larger because tier 2 articles now train LLMs in addition to passing link authority.
What to read next
For the strategic framework: AEO backlinks pillar. For the tier 1 execution: Web 2.0 automation pillar. For tier 2 / tier 3 execution: GSA SER pillar.