SEO's whole framework is simple. Technical, on-page, off-page — all of it in one thread.
If you genuinely understand this, you're already ahead of most people doing SEO.
Canonical: self-referencing by default. Point it at whatever page you want to rank.
Google claims it sometimes ignores canonicals. In practice, set them and they're respected.
In topic clusters, point supporting pages' canonicals at the hub. Concentrate the signal.
Hreflang isn't about duplicate-content penalties. There are none.
What it prevents is cannibalization: identical pages chasing identical keywords — and none of them ranking.
Speed is per page, not per site.
Every builder element — text block, empty container, carousel — slows the page. Plugins aren't the problem. Render-blocking and conflicts are. Load async and move on.
Google doesn't care how many 404s you have. Zero impact.
404 vs 410: 410 deindexes faster, but it burns the slug — and exact-match URLs still carry relevance value.
Pruning triggers a site-level reassessment on its own cycle anyway. Against that cycle, they're the same.
I've used plenty of 410s. Now I just 404.
301 is a merge. Only worth it when the merged page brings real links and traffic.
Most 301 situations are duplicate content that nothing links to. Usually: just 404.
On-page: exact-match slug. Exact-match H1. H2/H3 = variants of the main keyword.
That's relevance done. The rest is authority — and authority only comes after rankings and traffic. Real traffic, not a trickle.
Internal links don't split evenly.
What matters: the source page's traffic, relevance, rankings. And keep the count low. A page with strength 2 and two outlinks passes ~1 through each.
AI extracts content paragraph by paragraph — passage indexing all over again.
Everything under an H2 must fully answer that H2, so the passage stands alone.
And make your brand the subject of the sentence. Not "we". Not "I".
AI lifts whole sentences. If the subject is your brand, every extraction carries your brand. That's a contribution. Same rule off-site.
Product schema lifts CTR, and price fields get picked into AI results more easily.
Organization schema's sameAs/about = a warehouse: products, socials, markets, contacts in one place. Basic brand entity work.
Topic authority, plainly: the more pages, topics, and intents you rank within a topic, the stronger you are.
Hub page + articles covering every block — each one ranked. Not written at random.
Content is the big move.
Pull enough traffic curves: pure product and category pages almost never rise fast without outside force. Articles have a fast lane.
"SEO isn't just writing articles" — true. But articles are the most important early move.
A detail from a ranking site: keyword variants written into code, machine-only — which is hidden text, blackhat.
Then the same variants listed in small print, visible to users. Machine sees it, user sees it. No violation.
Not a recommendation. Just how far people go on details.
Off-page, first rule: DR, AS, every authority score — ignore them all. Useless.
Press releases buy presence, not rankings. More places, louder signal, deeper brand impression. Cheap = fewer outlets, expensive = Yahoo Finance tier.
Simple: money does it.
Backlinks you can buy today do basically nothing, however expensive.
Truly vertical, relevant links matter — and are almost never for sale. Outreach, skyscraper, guest posts: near-zero success rates. Don't sink your time there.
Programmatic SEO only works with unique data or real entities behind it. KFC, Apple — real stores, real addresses, real phone numbers.
Without unique data, it's mass content abuse. Spam.
The other valid shape: local. "SEO agency New York." Plumbers. Trades.
Modern SEO is mostly won off-site. SEO everywhere.
Your site is fuel — articles, products, tools. Ignition is traffic: your own rankings, plus social. Pick one or two platforms you're actually good at. Video → YouTube. Images → Pinterest. Writing → LinkedIn, X, Reddit.
Don't try to do them all.
The framework is simple. Doing it isn't.
Patience, care, execution, constant testing. Judgment is trained by doing. There is no SOP — different site, stage, market: different calls.
That's all of SEO.
I make a post about how Google's own documentation says that EEAT is not a ranking factor.
I even show a screenshot of Google's very own SEO Starter Guide.
... and I still have people telling me that if you don't have EEAT you can't rank in Google. 🤦♂️
This is your competition
Oh dear. Someone just reached out to tell me how Google could detect EEAT if they wanted to....
Because yeah, Google is going to trust "claims' and author bios....
And of course - no mention of YMYL..... because that destroys the EEAT narrative.
I can fabricate heuristics and suspend logic and enter delusion at any time on my own
Content vs Authority: SEO’s original sin
For the first 12 years after Google launched, SEO was about one hard problem:
getting authority to your content.
Then people gamed it, and somehow “all SEOs” got blamed.
Google went from pretending SEO didn’t exist…
to launching Google Webmasters…
to giving us GSC…
and Matt Cutts – still the clearest window into how Google really thinks.
Here’s the part nobody seems to say out loud anymore:
Every website is, by default, a content repository.
In social media, content is both the delivery van and the package.
In advertising, money is the van and content is the package.
In SEO, content is the package and authority is the delivery van.
But the Ivory Tower SEOs – and Google – quietly rewrote the story:
“Content is the van and the package.”
TechSEO and Content SEO tribes echoed it.
Now 100% of new SEOs think “writing content” is SEO.
No.
If your content has no delivery van, it’s not “search engine optimized.”
It’s just another box sitting in a warehouse, waiting for a truck that never comes.
Your SEO tool's competitor list are not your real competitors.
Every tool derives it the same way. Whoever ranks for the most of the same keywords you do.
That gives you a traffic neighbor. It does not give you the companies your buyer is choosing between.
Those are different lists and they overlap less than anyone expects.
The publisher with a thousand pages in your category shares hundreds of terms with you and has never once appeared on a shortlist against you. They are not competing with you. They are standing next to you.
Meanwhile the two names that came up on your last sales call might rank for almost nothing.
They are invisible to the tool and they are the only ones costing you deals.
Ask one question on your next five calls.
Who else are you looking at?
Write down what they say. That's your competitive set.
It won't match the SEO tool's list, and every gap analysis you've run against the tool's list was pointed at the wrong company.
Tu recrutes pour un job seo fortement orienté technique (donc logs, siloing, automatisation, volumétrie)
90% des SEO vont te dire qu'ils sont seo tech car ils font attention aux "Core Web Vitals / EEAT (page auteur + favicon) + j'utilise parfois Screaming Frog un petit peu en version gratuite"
Le niveau moyen de la profession c'est catastrophique, à une période où c'est si simple d'avoir un différenciant et de t'auto former, c'est dur de trouver des excuses
This is the actual definition of YMYL:
For example, some topics could significantly impact the health, financial stability, or safety of people, or the welfare or well-being of society
Source:https://t.co/Hv3pvtYmf3
In 2016, or even 2006, did SEOs not do MOFU or BOFU?
We’ve been doing it forever.
And who remembers what actually triggers Navboost? Clicks alone? No.
So who’s inventing new lies for people this time?
Part 2: For all the EEAT and Content Quality folks: are you saying AI is incapable of making up an authory- sounding bio?
If you can make claims - then AI can make claims and no - Google cannot validate them.
Just read this;
"Google admitted that it couldn't tell the difference between AI EEAT content and Human EEAT Content" said Google PR Liason, Tommy Sawyer - a well known Google Author.
There - I just did it.