Every few months someone declares SEO dead, usually right after Google ships a new feature that eats a bit more of the search results page. It's not dying. What's dying is the version of SEO that never actually worked — it just happened to survive because nobody was checking.
I've been doing this long enough to have opinions I'd defend in an argument, so here they are.
"Content is king" was always a half-truth
Everyone repeats it, almost nobody acts on the part that's actually hard. The full sentence should be: content is king if it's better than what's already ranking, for a query where being better actually gets rewarded. Most content teams optimize for the first half and skip the second entirely.
I've watched teams publish a genuinely well-researched 2,000-word guide, wait a month, and get furious that it's sitting at position 14. Nobody asked whether the query it targeted was one where depth mattered, or whether the top three results were already thorough enough that a fourth thorough article added nothing. Being good isn't the bar. Being meaningfully better than ten other decent options is the bar, and that's a much smaller target than most content calendars account for.
Link building as an activity is mostly theater now
I'll say the unpopular version: outreach campaigns built around "hey, loved your article, mind linking to mine?" emails are close to worthless in 2026, and most people running them privately know it. Response rates are in the low single digits, and the links that do land tend to be the kind that don't move anything, because they're transactional by nature — the linking site doesn't actually care about your content, they're doing you a favor or taking a fee.
What still works is almost the opposite of a campaign: build something specific enough that someone in your industry stumbles onto it while solving their own problem and links to it because it was genuinely the best thing available. That's slower, it's not schedulable in a content calendar, and it doesn't produce a tidy monthly report. It's also the only version I've seen consistently hold up over years instead of quietly decaying after an algorithm update.
AI Overviews didn't kill SEO — they killed the lazy middle
There's a real panic right now about AI-generated answers eating clicks, and some of that panic is justified. But look at what's actually disappearing from the click stream: the generic "what is X" content, the reworded Wikipedia summary, the listicle that exists purely to wrap a keyword in a template. That content was already low-value. AI Overviews just made the low value visible by doing the same job faster.
What isn't disappearing is content that requires a real opinion, a specific experience, or a decision that depends on context an AI answer can't fully capture — comparisons where the "right" answer depends on the reader's specific situation, analysis that takes a stance rather than surveying options neutrally, anything grounded in doing the actual thing rather than describing it secondhand. If your traffic dropped because an AI Overview now answers your query directly, the uncomfortable question worth asking is whether your page was ever doing more than what an AI Overview does.
E-E-A-T isn't a ranking factor, and treating it like a checklist misses the point
People talk about E-E-A-T like it's a technical requirement you can satisfy — add an author bio, get a few citations, done. It's not a factor you tick off. It's closer to a lens Google uses to evaluate everything else, especially for topics where being wrong has real consequences.
The mistake is optimizing for the signals of expertise instead of the substance. An author bio doesn't make a page trustworthy. A page that clearly reflects someone who's actually dealt with the problem does — and that shows up in specificity, in acknowledging trade-offs instead of pretending there's one right answer, in details that wouldn't occur to someone who only researched the topic secondhand. You can't shortcut that with a schema markup tag.
Technical SEO checklists are mostly a security blanket
I say this as someone who cares a lot about technical SEO: most checklists people run through — meta description length, header tag hierarchy, alt text on every image — matter far less than the panic around them suggests. They're not nothing, but they're table stakes, not differentiators. Nobody outranks a competitor because their meta description was 155 characters instead of 165.
The technical issues that actually move outcomes tend to be structural and unglamorous: crawl budget getting wasted on thousands of near-duplicate pages, internal linking that buries your best content four clicks deep, render-blocking scripts that delay content long enough for real users to bounce before Google even finishes indexing. Checklists feel productive because they're easy to complete. The stuff that matters is usually harder to find and less satisfying to fix.
Most "SEO strategy" is copying what already ranks, dressed up as strategy
This is the one that'll get me in trouble, but here it is: a huge amount of what passes for SEO strategy is looking at what's already ranking, reverse-engineering its structure, and producing something structurally similar with your own facts swapped in. That's not strategy, that's pattern matching, and it has a ceiling — you can get close to what's already working, but you can't outrank something by being a slightly different version of it.
The teams that actually break into a competitive space tend to do something the ranking pages aren't doing at all: a genuinely different angle, data nobody else compiled, an opinion the existing pages are too neutral to take. Copying the winner gets you into the pack. It rarely gets you out in front of it.
None of this is a formula, and that's sort of the point
I don't think there's a clean playbook here, and I'm suspicious of anyone who says there is. Search evaluates content the way people evaluate content, imperfectly and with a lot of contextual judgment, and it keeps getting better at spotting the difference between something built to rank and something built to actually help. The teams doing well right now aren't the ones with the most sophisticated checklist. They're the ones who stopped optimizing for the algorithm's approximation of quality and started just trying to be the best answer, and let the ranking be a side effect of that instead of the goal.
