Every SEO tactic has a fan club. Spend an hour on LinkedIn and you will be told that schema markup is the secret nobody talks about, that your Domain Authority is holding you back, and that shaving 200ms off your Largest Contentful Paint will unlock page one.
Most of that is noise. Some tactics genuinely compound, some are table stakes you fix once, at least one popular metric is not a ranking factor at all, and a new crop of "AI SEO" tactics has arrived with more confidence than evidence.
This page explains what each tactic actually is, then ranks it by return on effort in 2026, not by how often it gets talked about. Every tier was re-checked against Google's documentation on 5 October 2026.
The tier list#
| Tier | Tactic | Why it lands here |
|---|---|---|
| S | Topical authority | Compounds across every page you publish |
| S | Original data and research | The only asset competitors cannot copy |
| S | Internal linking | Free, fully under your control, chronically neglected |
| A | E-E-A-T signals | Not a score, but the thing every quality system approximates |
| A | Digital PR | The most durable way left to earn real links |
| B | Schema markup | Wins rich results, not rankings or AI citations |
| C | PageSpeed and Core Web Vitals | Real, small, and heavily over-weighted by developers |
| D | Domain Authority | Not a Google metric. A scoreboard, not a strategy |
| D | llms.txt | Google Search ignores it; no major engine has committed to using it |
| F | FAQ schema for rich results | The rich result no longer exists (gone since 7 May 2026) |
| F | Manufactured mentions and self-ranking listicles | Spam-policy territory now that the policies cover AI answers |
S tier: the compounding three#
Topical authority#
What it is: covering a subject so completely that a search engine has no reason to send the query anywhere else. Not one article on "crawl budget", but the full cluster: what it is, how to measure it, how to diagnose waste, how to fix it, and how it interacts with faceted navigation.
Why it is S tier: it is the closest thing modern SEO has to a flywheel. Each guide you add makes the surrounding guides more credible, deepens internal linking opportunities, and widens the set of queries you can plausibly answer. Google's ranking systems work mainly on individual pages but also use site-wide signals, and every major update since 2022 has rewarded sites that are consistently good on a subject.
The trap: topical authority is not "publish 200 thin posts on one theme". Scaled thin content is exactly what Google's scaled content abuse policy (March 2024) targets, and in May 2026 Google added that pages made for every query variation or AI "fan-out" query count too. Depth per page is the requirement, breadth is the multiplier.
Start with Topical Authority.
Original data and research#
What it is: publishing something only you could publish. A survey of 500 practitioners. A teardown of 10,000 SERPs. Benchmarks from your own product telemetry. Anything where you are the primary source rather than a summary of other people's summaries.
Why it is S tier, and more so every year: it is the one thing that survives commoditization. Explanatory content is now trivially generated at scale, which means the marginal value of the 400th "what is schema markup" article is approximately zero. A number nobody else has cannot be regenerated. It earns links without asking, gets cited in AI answers, and gives digital PR something real to pitch. Google's 2026 guidance on AI features makes the same point in its own words: focus on unique, "non-commodity" content.
The trap: it is genuinely expensive, and a weak dataset is worse than none. If your sample is 40 responses from your own newsletter, say so plainly rather than dressing it up as an industry study.
Internal linking#
What it is: the links between your own pages. Which pages you link to, from where, and with what anchor text.
Why it is S tier: it is the highest-leverage tactic almost nobody finishes. It costs nothing, requires no outreach, no approval, and no budget. It distributes authority from your strong pages to your weak ones, tells search engines which pages you consider important, establishes topical relationships, and is often the difference between a page being crawled monthly and daily.
The trap: most sites link generously in navigation and templates, then almost never contextually from body copy. Body links carry more weight and are the ones actually missing.
Start with Internal Linking.
A tier: high value, higher cost#
E-E-A-T#
What it is: Experience, Expertise, Authoritativeness and Trust. It comes from Google's Search Quality Rater Guidelines, the manual that human raters use to score result quality.
The nuance that matters: E-E-A-T is not a score in the algorithm. There is no E-E-A-T number attached to your domain. Raters do not directly change rankings. What they do is evaluate whether algorithm changes moved results in the right direction, which means Google builds systems that approximate what raters reward. So it is not a ranking factor, and optimizing for it still works.
Practically: real author bylines with real credentials, citations to primary sources, clear ownership and contact information, visible update dates, and first-hand experience stated plainly. The Experience E, added in 2022, is the easiest to earn and the most neglected: say that you actually used the thing.
Why A and not S: it is close to mandatory for YMYL topics (health, finance, legal, safety) where it moves to S. For a developer tools blog it matters considerably less.
Start with E-E-A-T and Quality Guidelines.
Digital PR#
What it is: earning coverage and links from real publications by giving journalists something worth writing about. The modern descendant of link building, minus the directories and guest post farms.
Why it is A tier: links still matter, and this is the most durable way left to get good ones. A link from a national title carries authority that no amount of outreach templating replicates. It also builds brand search volume, which is itself a meaningful signal.
The trap: it is slow, expensive, and has a high failure rate. Most campaigns land nothing. It also depends entirely on having something to pitch, which is why it pairs with original research and fails without it. The shortcut version, buying "mentions" or placements to look popular to AI systems, is in F tier below.
Start with Digital PR and Brand Signals.
B tier: do it once, do it right#
Schema markup#
What it is: structured data, usually JSON-LD, that labels what a page contains. This is an Article, published on this date, by this author. This is a Product, priced at this, with this rating.
Why it is B and not higher: Google is explicit that structured data is not a ranking factor. Adding Article schema will not lift you a position. What it does buy is real: eligibility for the rich results Google still shows (review stars, products, recipes, events, videos, desktop breadcrumbs), which can lift click-through rate on positions you already hold. The list keeps shrinking, though: FAQ rich results ended in May 2026 and HowTo in 2023.
What it doesn't buy: a ticket into AI answers. Google says there is no special schema.org markup needed for AI Overviews or AI Mode, and lists "overfocusing on structured data" among the AI-era tactics to ignore, while still recommending it for rich results.
Why it is not lower: it is cheap, one-time, and low-risk. Most sites should implement Organization, Article and BreadcrumbList and stop thinking about it.
The trap: marking up things that are not on the page. Fake review markup and inflated aggregate ratings are manual-action territory, and Google added a rule against fake and undisclosed incentivized reviews in July 2026.
Start with Structured Data.
C tier: real, but you are probably over-indexed on it#
PageSpeed and Core Web Vitals#
What it is: how fast and how stable your page feels. Measured through three metrics: Largest Contentful Paint (loading), Interaction to Next Paint (responsiveness, which replaced First Input Delay in March 2024), and Cumulative Layout Shift (visual stability).
Why it is C tier for rankings: page experience is a genuine signal and a genuine tiebreaker between otherwise comparable results. It is also small. The thresholds (LCP 2.5s, INP 200ms, CLS 0.1 at the 75th percentile) have not changed since INP arrived, whatever 2026 blog posts claim. Google has said repeatedly that relevance outweighs it, and the SERPs agree: slow pages with the best answer routinely outrank fast pages with mediocre ones.
Why developers get this wrong: it looks like engineering, has a numeric score, and produces a satisfying green circle. Why you should still do it: the conversion and UX case is far stronger than the ranking case. Fix it for revenue, not for position.
Start with Core Web Vitals.
D tier: stop optimizing for this#
Domain Authority#
What it is: a 1-100 score predicting how well a domain might rank. The name usually refers to Moz's metric; Ahrefs has Domain Rating, Semrush has Authority Score. All are third-party estimates built mostly from backlink profiles.
Why it is D tier: Google does not use it. It is not a Google metric, it is not an input to Google's systems, and Google has stated they have no such single "domain authority" score. These tools are estimating an outcome from the outside; treating the estimate as the target inverts cause and effect.
What actually goes wrong: teams set "raise DR to 60" as a goal, then buy links to move a number that no search engine reads. The score goes up. The traffic does not.
Where it is still useful: as a rough comparative filter. Is this prospect roughly in our league, is this competitor plausibly beatable, did our link profile grow this quarter. Directional, comparative, and never a KPI.
llms.txt#
What it is: a proposed Markdown file at /llms.txt that summarises a site and links its key pages for language models.
Why it is D tier: Google's docs say plainly (June 2026) that Google Search doesn't use llms.txt and having one will "neither harm nor help" your visibility. No major answer engine has publicly committed to using it for ranking or citation. Chrome's Lighthouse added an llms.txt check to its experimental Agentic Browsing category in May 2026, but that is a tooling check for AI agents, not a ranking signal.
Where it is still reasonable: as a cheap convenience for agents and developer tools navigating a site you already serve (Rankyl serves one). Spend an hour, not a sprint.
F tier: stop doing these#
FAQ schema for rich results#
FAQ rich results were limited to government and health sites in 2023, stopped showing for everyone on 7 May 2026, and Google removed the documentation on 15 June 2026. Existing FAQPage markup is harmless and needn't be ripped out, but adding it for a SERP feature is effort spent on something that no longer exists. A visible FAQ section that answers real questions is still fine content. See Structured Data.
Manufactured mentions and self-ranking listicles#
The 2026 "GEO" shortcut: pay for placements, seed fake forum posts, or publish "best X" lists where your own product is #1, hoping AI answers repeat them. Google's AI optimization guide (May 2026) says seeking inauthentic mentions "isn't as helpful as it might seem", its spam policies now cover attempts to manipulate generative AI responses, and a Google spokesperson said in April 2026 that it is "aware of the low-quality listicle content". It is the link-buying of the AI era, and likely to end the same way. Earn real coverage instead: Digital PR & Brand Signals and Reddit, Forums & Communities.
What changed for 2026#
The tier list moved for one reason: answers moved. AI Overviews had over 2.5 billion monthly users and AI Mode over 1 billion by May 2026 (Google's figures). When a meaningful share of queries are resolved inside an AI answer, being the source that gets cited matters more than owning the fourth blue link. And Google says those AI features are rooted in its core ranking systems, so the classic fundamentals carry over.
That shift rewards the top of this list and punishes the bottom. Original research gains the most, because you cannot be a primary source by summarizing. Topical authority and E-E-A-T gain, because consistent, credible, named sources are easier for any system to trust. Domain Authority loses further: never read by Google, certainly not read by a language model. Schema holds steady, but only for rich results. And the "AI-only" tactics (llms.txt, chunking, manufactured mentions) sit at the bottom because Google's own documentation says they don't help in Google Search.
See Generative Engine Optimization and SEO for AI Mode & AI Overviews for how citation actually gets earned.
If you only have ten hours#
In order:
- Three hours: internal linking. Audit your top 20 pages. Add contextual body links from your strongest pages to the ones you want to rank. Free, immediate, and almost always incomplete.
- Two hours: E-E-A-T basics. Real byline, real author page, contact details, visible update dates. Cheap, permanent.
- Two hours: schema.
Organization,Article,BreadcrumbList. Validate it. Move on. - Three hours: pick one topical gap and fill it properly. One genuinely complete guide beats five shallow ones.
Notice what is not on that list: buying links to lift a third-party score, chasing your Lighthouse number from 88 to 96, writing an llms.txt "for Google", and adding FAQ schema.
Key takeaways#
- Rank tactics by return on effort, not by how much attention they get. Internal linking is the most underused free lever.
- Original research is the only content type that cannot be commoditized, and it is what AI systems cite.
- E-E-A-T and Domain Authority are both not Google scores; the first still works because Google approximates it, the second is a vendor metric. Never make DA a KPI.
- Schema wins rich results, not rankings or AI citations. Core Web Vitals are worth fixing for users and revenue more than position.
- llms.txt is ignored by Google Search, FAQ rich results are gone, and manufactured mentions are a liability.