8 SEO Tactics I Stopped Running Between 2022 and 2026

TL;DR: SEO isn't dying. A specific 2020-era playbook is, and most postmortems duck the specifics. Eight tactics lost ground between 2022 and 2026: exact-match anchors at scale, thin programmatic SEO, generic FAQ schema everywhere, bought guest posts, position-1 chasing on AI Overview queries, bulk LLM content refreshes, second-tier directory citations, and volume-as-authority thinking. Not all of them died cleanly, and one is messier than the rest. But the broad pattern holds: cheap, scaled, placement-volume work lost ground; depth, first-party data, and earned mentions gained it. The 2026 audit fits on a five-question checklist.

Last week I opened a tab on "is SEO dying," the same query I've watched resurface every 18 months for a decade, and the front page of Google had eight different blogs telling me the same thing. SEO is still alive. Don't panic. Adapt or die. Read their newsletter.

None of them named what specifically stopped working. Which tactic, in which year, after which Google update. I ran most of these for clients between 2018 and 2021, had to stop between 2022 and 2024, and I want to be precise about which ones broke, which ones just faded, and what replaced them. The discipline didn't die. A playbook did, unevenly.

SEO tactics that stopped working in 2026 — hero illustration — A 2020 SEO toolkit photographed in 2026. Most of these levers don't move the needle anymore.

Exact-Match Anchor Text at Scale

From roughly 2014 to 2020, you could move a page by negotiating links with the exact target keyword in the anchor. Cheap, repeatable, scalable. The whole "guest post on a niche blog with 'best CRM software' in the anchor" industry was built on this lever.

It didn't break on a single date. Google's link-quality work through Penguin (2012-2016) and SpamBrain (2018-onwards) reduced the lift first, gradually, and the decline was never clean enough to point at one update. Then John Mueller said the quiet part out loud, not in one memorable quote, but as a position he held consistently across years of Google Search Central Office Hours through 2024: low-quality or paid anchor text isn't penalized, it's ignored. You're paying for a link Google already discounted to near zero. I say "near" deliberately, because in a few low-competition niches I've watched these links still nudge things, which is exactly why the tactic refuses to die quietly in the agency decks that sell it.

What replaced it, for the sites that adapted, is internal linking discipline. Patrick Stox at Ahrefs has argued consistently (most visibly off his 2024 talk auditing over a million sites) that poor internal linking is one of the highest-impact problems most sites have, ahead of the next batch of external links you might chase. It tracks with what I see in audits: a B2B SaaS with 200 strong external links and an incoherent internal-link graph almost always underperforms a competitor with 100 external links and a clean topical structure.

The replacement bundle:

Internal linking organized into topical silos with intentional anchor variety. See our guide to auditing content silos.
Earned brand mentions in trade communities. Unlinked mentions count toward entity recognition even when they don't pass PageRank.
The occasional high-quality external link earned by being citable, not by being purchased.

Thin Programmatic SEO Landing Pages

If you ran an affiliate or comparison site between 2017 and 2022, you almost certainly had programmatic templates. One template, one data source, ten thousand URLs. "Best running shoes under $100 for flat feet in Portland" and ten thousand near-permutations.

Then came the Helpful Content Update, in three big rounds: September 2023, March 2024, and August 2024. Plus the SpamBrain refinements running through 2025. The pattern was stark. Sites with thousands of templated pages built on Amazon API data, without first-party testing or original commentary, lost most of their organic traffic in a single update cycle. Paul Teitelman's six-month study found nearly half the sites he tracked lost more than 90% between December 2023 and August 2024.

Traffic decay curve for a programmatic affiliate site hit by HCU — A composite traffic decay curve typical of HCU-impacted programmatic affiliate sites, 2023-2024.

The case study most people in SEO know is HouseFresh, the air-purifier review site whose founder Gisele Navarro published a detailed early-2024 postmortem on how their organic traffic collapsed while larger sites with thinner programmatic pages kept their rankings. It was specific: the affected URLs, the competitor pages, the API-data-only patterns that survived (briefly) before the next HCU round caught them too. HouseFresh lost roughly 91% of its organic traffic by the March 2024 update. Ninety-one percent.

What replaced programmatic-at-scale isn't "stop using templates." It's "templates are fine if and only if they ship first-party data the user can't get elsewhere." A template wrapped around your own measurement, your own teardown, your own pricing analysis is a deep page in a template shell. A template wrapped around Amazon's API is filler. Google can tell the difference now, with more confidence in 2026 than I'd have claimed in 2022, when the signal was noisier and some thin sites were still hanging on.

If you operate a programmatic surface, the 2026 question is concrete. What does each generated page contain that a reader couldn't reconstruct from public APIs in five minutes?

Generic FAQ Schema on Every Page

From 2018 through 2022, appending FAQPage schema to every article was a free CTR boost. The FAQ rich-result expanded your listing height in the SERP, pushed competitors lower, and gave you accordion real estate. So every blog post got an FAQ section, often with questions invented for the schema rather than questions users actually asked.

In August 2023, Google restricted FAQ rich results to authoritative government and health websites. The accordion disappeared from the SERP for almost everyone overnight. Then came the part that makes this section fully past-tense: as of May 7, 2026, Google deprecated FAQ rich results entirely, for every site, including the government and health sites that kept them after 2023. There is no FAQ rich result left to earn anywhere.

What replaced it isn't more schema, it's better-targeted schema. FAQ markup is still valid, and there are still reasons to use it: answer engines parse structured Q&A, and your own on-page accordion may render it. It just no longer earns a Google SERP feature. For other content shapes, the schema that earns rich results in 2026 is more often Article, HowTo, Product, or Review. Match the schema to what the page actually is, and keep it accurate; bad markup quietly suppresses the feature rather than earning it.

Buying Guest-Post Links from "SEO Guest Post" Lists

The 2018-2021 link-building industry ran on placement lists. You'd pay $50-300 for a link in an article on one of the same 200 sites everyone else was using. Some lists overlapped 80% between agencies. The product wasn't editorial coverage. It was a slot in a circular link economy.

Two things broke this. First, Google's Search Liaison and John Mueller clarified through 2023 and 2024 that guest posts written to acquire links violate Google's spam policies, and the March 2024 site-reputation-abuse enforcement made it concrete, penalizing or deindexing more than 400 sites for hosting low-quality SEO-driven content. Second, and this part is my own ledger rather than a published study, by the time I stopped buying placements most produced no measurable ranking movement at all.

Comparison of bought guest-post links versus earned mentions — Composite outcome distribution for paid guest-post placements compared with earned trade mentions, 2023-2024 industry data.

I bought guest-post packages for clients in 2019. By 2022 the placements were still landing, the links still appearing, and rankings weren't moving. The supply chain was intact and the product was inert. I can't give you a clean percentage, only the pattern across the accounts I tracked, which is exactly why I won't dress it up as research.

What replaced bought guest posts isn't "stop building links." It's earned mentions in places that matter to your audience:

Trade press coverage of something you actually built or said.
Podcast appearances on shows your buyers listen to. Unlinked mentions count toward entity recognition.
Credible community discussion. Reddit, Hacker News, niche Slacks, industry Discords. The kind that develops organically because your work is useful.
Original research that other operators cite because it answers a question they had.

This is harder. It's also durable. We covered the discipline side of this in ethical SEO practices in 2026.

Chasing Position 1 for AI-Overview Queries

By 2025, position-1 organic CTR fell off a cliff on queries that triggered an AI Overview. How far depends on who's counting and over what window. Ahrefs measured the top result dropping from 7.3% to 2.6% on AIO queries across a year, while Seer Interactive put the drop closer to 60% in their late-2025 update. Call it a 35-60% hit depending on query type and study period. The full data set and the citation mechanics live in our deep-dive on optimizing for AI Overview citations.

Position 1 organic CTR comparison with and without AI Overview present — Composite CTR comparison for organic position 1 on queries with and without an AI Overview, based on aggregated 2025 reporting (Ahrefs, Seer Interactive, BrightEdge).

What replaced position-1 chasing is being the cited source inside the AI Overview itself. The mechanics aren't magic: comprehensive entity coverage, clean structured data, answer blocks that summarize the topic in the first two paragraphs, a named author or organization with verifiable credentials, and a clear topical position the page is organized around. We cover the operating model in ask-engine optimization and the citation mechanics in generative-engine optimization and mentioning.

The blunt version. For an informational query that triggers an AIO, "rank position 1" is a less useful goal than "be one of the sources the AIO cites." Different optimization, different KPI. How many sources get cited varies by query and shifts as Google tunes the feature, but the directional point holds.

Bulk LLM-Driven Content Refreshes

This one is recent and ongoing, and honestly the least settled on this list. From late 2022 through 2024, a common content-ops play was straightforward. Take your old top-traffic blog posts, feed each one to an LLM with a "make this better, modernize the examples, expand the depth" prompt, and republish, at scale, across hundreds of URLs.

I don't have a clean published statistic, and I'd rather say so than invent one. What I have is the pattern in the bulk-refresh projects I've reviewed: the majority of fully LLM-rewritten posts lose traffic against their pre-refresh baseline. The why is consistent. Bulk rewrites flatten voice, add hedged filler, restructure sections away from what was actually ranking, and swap specific claims for generic ones. The model averages out the page. But it's not universal, which keeps me cautious about declaring it dead: I've seen a handful of rewrites on genuinely stale, thin posts come out ahead, because the bar was so low that even an averaged-out page beat the original. So the tactic isn't dead so much as a bad default. (Animalz make the adjacent point that refresh cadence beats refresh volume, but that's a different finding, so I'll keep the two separate.)

Refresh-method outcome distribution: bulk LLM versus human-edited — Composite outcome bucket distribution for content refresh methods, based on field experience across reviewed bulk-refresh projects.

What reliably works is targeted, human-edited refresh on the 10-20% of pages that drive 80% of your organic traffic. The decision isn't "should I refresh." It's "which page, in which section, with what specific update, by which named editor." We laid out the decision tree in our content refresh strategy guide. An LLM is a fine tool inside a refresh, good for finding stale stats or suggesting reorderings. It's a bad tool for the act of refreshing itself, when "refresh" means "rewrite the whole post and publish."

A Few Tactics on the Edge

Three more that have decayed enough to flag but don't rate a full section, and one of them genuinely isn't dead.

Second-tier directory citations for local SEO. This is the messy one. The list of "150 directories your business should be on" is mostly noise for traditional ranking in 2026. Whitespark's Local Search Ranking Factors research showed citation count flattening as a ranking signal years ago, with Google Business Profile signals gaining relative weight. So far, so dead. But here's the wrinkle I won't tidy up: citations have quietly come back as a factor for AI search visibility, where models lean on consistent, structured business data across sources. So the long tail of 150 is still dead weight, while accuracy on the top 20 matters more than it did in 2020, not less. I genuinely don't know where the line settles, because the AI-visibility angle is moving every quarter. What replaced raw citation count is GBP completeness, first-party reviews on the platforms your buyers use, and locally-relevant earned mentions. Clean the top 20; skip the next 150.

Volume of content as a topical-authority signal. "Publish more content on the topic" got coupled, in some agencies, with the idea that the count itself signaled authority. The Helpful Content Update made depth-per-URL the operative signal; a single thorough page typically outperforms five thin ones on the same intent. I want to be careful here, because the popular shorthand that "HCU survivors had higher human-authored ratios" sounds true but doesn't survive the data. Zyppy's own 50-site case study found no statistically significant relationship between author signals and traffic outcomes. So I'll make the narrower claim I can stand behind: depth per URL beat volume of thin URLs. The author-box folklore is not something the data supports.

Indiscriminate redirect chains for "consolidation." 301-everything-to-your-money-page used to be a recovery tactic; it now correlates with manual review attention. Surgical 301s based on actual page overlap, kept short, are what work. If you've inherited a redirect graph, audit before adding to it.

What This Means for Your 2026 SEO Plan

Stand back and the pattern is mostly consistent. The tactics that lost ground were cheap, scaled, and placement-volume-shaped. They faded as the SERP started rewarding depth, first-party data, and earned mentions. The exceptions, the low-competition anchor that still nudges, the rewrite that beats a truly thin original, the citation that now feeds AI visibility, are why "dead" is the wrong word for some of these. "Demoted to a bad default" is closer.

Here's the five-question audit I run on a client stack now. It takes about an hour.

How many of your inbound links are exact-match anchors from low-DR domains? If more than 20%, that bucket is doing little for you and may be drag.
How many of your indexed URLs are programmatic templates without first-party data inside? If your "deep page" coverage is templated and shallow, plan to either deepen or noindex.
Where do you have FAQ schema that was earning rich results? It earns none now. Keep it only where it serves answer engines or your own on-page accordion, and strip the rest.
What percentage of your link-building line item is paid guest posts? Reallocate at least half of that budget to original research, podcast pitching, and trade-press outreach.
What percentage of your most recent "content refresh" was LLM-driven without human editing? Flag those pages for re-refresh by a named editor, starting with the ones that lost traffic.

Two larger shifts to plan around. First, owned distribution is back in fashion for a reason; we covered the compounding-channel argument in SEO as an owned distribution channel. Second, the operating model matters more than the tactics. Building an SEO system that runs without you is the meta-skill that ties everything above into something a team can actually execute.

Here's the 2026 summary table I put on a slide for clients now:

Stopped working (2020-era)	When it shifted	Replaced by (2026)
Exact-match anchor text at scale	2014-2024, gradual (Penguin and Mueller's consistent position)	Internal-linking discipline plus earned brand mentions
Thin programmatic SEO pages	2023-2024 (HCU rounds)	Templates wrapped around first-party data
Generic FAQ schema everywhere	August 2023 restriction, then full deprecation May 2026	Targeted FAQ for answer engines and accurate schema types per page shape
Bought guest-post links	2023-2024 (spam-policy clarifications and 2024 enforcement)	Earned trade mentions, podcasts, community discussion
Position-1 chasing on AIO queries	2024-2025 (AI Overview rollout)	Being the cited source through AEO and GEO mechanics
Bulk LLM content refreshes	2024 onward (field experience; a bad default, not always fatal)	Targeted human-edited refresh on the top 10-20% of URLs
Second-tier directory citations	Gradual 2022-2024, partly reversing for AI visibility	GBP completeness, first-party reviews, earned local mentions
Volume-of-content as authority signal	2023-2024 (HCU)	Depth-per-URL on fewer pages, deeply cross-linked

None of this means SEO is harder. It means the cheap arbitrage is closed and the work is the work. If you're inheriting a stack that leans on these tactics, the audit pass is the first move. The SEOJuice site audit will surface most of the technical surface in one pass. The strategic questions above are still yours to answer, but the inventory part doesn't need to be manual.

Frequently Asked Questions

Is SEO dying in 2026?

No. The 2020 playbook is. The discipline of helping search engines understand what your page is, who wrote it, and why a reader should care, that is more important than it was five years ago, not less. The tactics that died were the cheap, scaled, placement-volume ones.

Does exact-match anchor text still help in 2026?

Mostly no. John Mueller's position has been consistent across years of Office Hours through 2024: low-authority or paid exact-match anchor text is ignored, not penalized. There's little ranking lift left, though I've seen it still nudge things in a few low-competition niches. The replacement is internal linking discipline and earned brand mentions.

What replaced guest posting for SEO link-building?

Earned mentions. Trade press coverage of something you built, podcast appearances on shows your buyers listen to, credible community discussion (Reddit, niche Slacks, industry Discords), and original research that other operators cite because it answers a question they had. Harder to acquire, far more durable.

Do FAQ rich results still work in 2026?

No. Google restricted FAQ rich results to authoritative government and health sites in August 2023, then fully deprecated them for all sites as of May 7, 2026. There is no FAQ SERP feature left to earn. FAQ schema is still valid markup and useful for answer engines, but it no longer renders an accordion in Google Search for anyone.

Are AI Overviews killing organic traffic for everyone?

Mostly for informational queries. The top-funnel "what is X" and "how does Y work" intents. Commercial and navigational queries are less affected because the user usually wants to land on a specific site or product page. Your AIO exposure depends on your query mix more than your domain.

The URL of this article changed. What happened?

You're reading the rewrite of what used to live at /blog/is-seo-really-dying/. The old slug now redirects here. The original article asked the wrong question and ranked for nothing. The new one names the tactics that actually stopped working. The redirect keeps any existing backlinks pointing at the right answer.

Run a free SEO audit to see how many of these eight tactics are still sitting in your own stack.

Our powerful suite of automation tools for SEO

Learn, discover, and get inspired by our content