Forcing the keyword into the first 100 words

"Put your exact keyword in the first sentence — early placement ranks."

This is cargo-cult on-page.

There's no documented 'first 100 words' boost. What there is: the failure mode of writers torturing an intro to front-load a phrase, producing the stilted 'In this article about [exact keyword], we will discuss [exact keyword]...' opener that signals low quality more reliably than almost anything else.

Here's what actually happens: you trade a real benefit — a clear, useful opening that answers the query fast — for a superstition. The page that wins the snippet isn't the one that says the keyword first; it's the one whose first paragraph answers the question first.

Fix: lead with the answer, in plain language. The relevant terms show up naturally when you actually address the topic. Stop engineering the intro around word position and engineer it around the reader's first question. (Robots stopped counting word #1 a long time ago.)

The slow-bleed redirect chain from URL 'cleanups'

"We tidy our URLs every redesign — old links still work, we redirect them."

Until the chain eats them.

Each redesign adds a hop: /old → /v2 → /new → /final. Every link still 'works,' so nobody notices the mistake — but each extra hop loses a little signal, slows crawl, and after enough chains some bots stop following.

Here's what actually happens: three redesigns later, a page acquires a four-hop chain, the canonical points at hop two, and the sitemap lists hop three. Now your strongest backlinks dribble through four 301s to land on a URL your own systems disagree about.

Fix: after any URL change, audit redirects and flatten chains — every old URL should hop once, directly to the final destination. Update the canonical and the sitemap to the final URL too. One hop, no detours. (A redirect that works is not the same as a redirect that's efficient.)

The templated-title duplicate problem

"We auto-generate titles from a template — consistent and scalable."

Consistent, yes. Distinct, no.

The failure mode of {Product} | {Category} | {Brand} at scale: pages whose only difference is a product name, producing near-identical titles across hundreds of URLs. Google sees a wall of boilerplate, picks a few to rewrite, and treats the rest as low-differentiation.

Here's what actually happens on big sites: the template looks fine for the ten pages you check and breaks on the thousands you don't. Two products with similar names get titles that differ by one word, both fighting for the same SERP, neither distinct enough to win.

Fix: bake real differentiation into the template — pull in a spec, a price, a use-case modifier, a unique attribute from the data. {Product} {KeySpec} | {Category}. The template should generate variety from your data, not stamp the same shape on everything. Scale is no excuse for sameness.

ЕЮ Иванов продолжает кошмарить АффПапу, конторку, которая накинула говна на вентилятор этим летом. Тогда в AffPapa не знали, с каким говном идут бодаться, поэтому заслуженно проиграли. 😏

На этот раз ЕЮ зарегал товарный знак AffPapa — совсем скоро имя компании будет официально принадлежать ему. Чтобы убедиться в трушности мува, переходим по ссыл-Очке и вводим серийный номер: 2026793242. Там видим, что заявка на регистрацию подана лично Евгением Юрьичем.

Всё это выглядит забавно, но давайте не забывать, в какой сфере мы работаем и что реально может произойти с жирным троллем за воровство нейминга. Впрочем, толстому не привыкать отхватывать пиздов за проделки в интернете, поэтому ждем очередную фотку разбитого ебала и длинный пост с извинениями. 😏😏😏

В арбитраже денег нет 💵

Still relying on rel=next/prev

"We mark up pagination with rel=next/prev so Google understands the series."

Google stopped using that in 2019.

They announced they hadn't used rel=next/prev for indexing in years — and people are still shipping it as their entire pagination strategy, then wondering why deep pages don't get crawled or why page 2+ products never index.

Here's what actually happens: the markup does nothing, so the real failure underneath stays unsolved — paginated pages with thin, near-duplicate content, canonicals pointing every page back to page one (which deindexes pages 2+ and the products only listed there), and no other path for crawlers to reach deep items.

Fix: don't lean on a dead signal. Make each paginated page self-canonical and crawlable, ensure products are reachable from somewhere other than deep pagination (category links, sitemaps), and keep the on-page content meaningful per page. Keep rel=next/prev if you want — for browsers — but it's not an SEO plan.

The meta keywords tag, in 2026

"Filling the meta keywords tag — every little helps."

It doesn't help. It can hurt.

Google has publicly ignored the meta keywords tag since 2009. So the 'every little helps' instinct buys you exactly zero ranking benefit. But here's the actual mistake: you're publishing your target keyword list, in plain HTML, to anyone who hits View Source — including competitors who'll happily read your strategy off the page.

Here's what actually happens: a team keeps a legacy field in their CMS, dutifully fills it, hands competitors a free keyword map, and gains nothing. The only entity reading that tag is the one you least want reading it.

Fix: delete the field, or leave it blank. Spend the effort on the title, the H1, and the body copy — the things that are actually read. (A tag that helps nobody but your rivals is not 'every little helps.' It's a leak.)

Padding to hit a word count

"Top results average 2,000 words, so make every page 2,000 words."

You've confused correlation with a quota.

The data is real — long content correlates with ranking. The mistake is reading a correlation as a target and padding thin topics to hit it. Comprehensive pages tend to be long because the topic demanded length, not because the author was filling a meter. Reverse the causality and you get bloat.

Here's what actually happens: a query that deserves a 400-word direct answer gets stretched to 2,000 with throat-clearing, restated points, and a 'what is' section nobody asked for. The padding pushes the actual answer below the fold, raises bounce, and dilutes the relevance you started with.

Fix: length should be an output of completeness, not an input. Cover the topic fully — every real subtopic, every reasonable follow-up question — then stop. If that's 600 words, it's 600 words. A tight page that answers fully beats a long one that stalls. (Nobody ever ranked for endurance.)

"Screaming Frog and Sitebulb find the same title problems."

No.

They crawl the same tags, but they flag different things. Screaming Frog hands you a raw column of pixel widths and truncation lengths and lets you sort. Sitebulb hands you a prioritized hint list that already decided what's "too long" for you.

That's the actual tradeoff, and it matters. If you're auditing 50,000 URLs and want to write your own rules — title-over-580px, missing keyword, duplicated across templates — Frog's flat export wins, because you control the threshold. If you're handing a report to a client who won't read a spreadsheet, Sitebulb's opinionated scoring wins.

The mistake is treating "it found fewer issues" as "the page is fine." It found fewer issues because someone else set the cutoff.

(Frog's default 561px truncation guess is also just a guess. SERP truncation moved years ago and nobody updated the folklore.)

"Ahrefs shows my internal link structure, so I don't need anything else."

No.

Ahrefs shows the internal links it crawled and chose to report, weighted by its own index priorities. Search Console's Links report shows what Google actually counted as internal links to each URL. Those are two different questions, and people conflate them constantly.

Use Ahrefs (or any crawler) when you're planning — finding orphans, mapping anchor distribution, seeing the hub-and-spoke before you build it. It sees the whole site at once, including pages Google hasn't bothered indexing.

Use GSC's Links report when you're diagnosing — "why isn't this page getting equity" — because it reflects Google's actual graph, including links it dropped from JS it didn't render.

The gap between the two lists is the interesting part. A page with 80 crawler-visible internal links and 3 in GSC has a rendering problem, not a linking problem.

(Yes, this means your beautiful mega-menu might count for almost nothing.)

Network picks — SEO & search:

— @OverviewHotTake — Strong, unfiltered opinions on AI Overviews and generative search —…
— @MapPackDiaries — Real local-SEO case studies told as stories — the dentist who tripled…
— @CartAndCrawl — Your e-commerce SEO questions, answered straight — faceted nav,…
— @StackAndCompare — Honest, hands-on SEO tool comparisons — Ahrefs vs Semrush vs the…