Splitting one bloated page into 3 ended cannibalization and added 5,200 clicks

One 4,800-word page was ranking for three distinct intents and competing with itself in Search Console.

— Queries the page ranked for: 140
— Queries where two of our own URLs appeared: 38
— Avg position before split: 14
— Avg position after split: ▇▇▇▇▇▇ 6

We clustered the 140 queries by intent, found three clean groups, and broke the page into three URLs with internal links between them.

Clicks across the trio rose from 3,100 to 8,300/mo over two months. The original URL kept its strongest cluster.

So what: a page ranking for too many intents is usually under-ranking for all of them.

Benchmark of the week: pages competing with themselves on 25%+ of queries average a 5-position penalty versus single-intent pages in our set.

Meta ограничивает расходы на токены для сотрудников

Meta ввела внутренние лимиты на использование ИИ из-за резкого роста расходов: в 2026 году только на сотрудников заложены миллиарды долларов, а общий бюджет на ИИ-инфраструктуру оценивается в 130–145 млрд. Вывод простой: даже у Big Tech ИИ перестал быть бесплатной игрушкой и требует жёсткого контроля затрат.

➡️ Читайте на сайте: https://aff.top/blog/meta-ogranichivaet-raskhody-na-tokeny-dlia-sotrudnikov

🧠 Ещё больше инсайтов → в канале AFF.top

Claude Cowork, Claude Design объединили в один Claude

➡️ Читайте на сайте: https://aff.top/blog/claude-cowork-claude-design-obedinili-v-odin-claude

🧠 Ещё больше инсайтов → в канале AFF.top

Re-prioritizing by impression floor, not volume, doubled win rate

We stopped picking targets by tool volume and started picking by existing Search Console impressions (queries we already showed for but ranked 11-20).

— Candidate queries on page 2: ▇▇▇ 640
— Avg tool-reported volume: 1,200/mo
— Pages optimized: 30
— Page-1 breakthrough rate: 61% vs 28% on cold targets

These were queries Google already associated with the site. Pushing a position-13 page to position 8 is far cheaper than ranking a cold keyword from scratch.

Thirty edits, mostly H2 additions and internal links, no new content from zero.

So what: your highest-probability targets are the ones you already rank 11-20 for. Volume tools never surface them; impression data does.

Benchmark of the week: page-2 queries convert to page-1 at 2.2x the rate of never-ranked targets with equal effort.

Internal site-search logs out-convert keyword-tool long-tail 2x
For long-tail discovery, external tools compete with your own data. We compared converting terms from a keyword tool vs your site's internal search logs.
— Internal search terms: ▇▇▇▇▇▇▇ lower volume but 2.1x conversion rate (real users, real intent, zero estimation)
— Keyword-tool long-tail: ▇▇▇▇ broader reach, more top-of-funnel
— 47% of converting internal-search terms had no volume reading in the external tool at all
— Internal logs only exist where you already have traffic
So what: mine internal search and on-site query logs before buying long-tail lists. Those zero-volume terms are pre-qualified demand the tools literally cannot see.
Benchmark of the week: roughly 45% of high-converting internal-search queries register as 'no data' in mainstream volume tools.

Microsoft планирует вставлять рекламу в игры

➡️ Читайте на сайте: https://aff.top/blog/microsoft-planiruet-vstavliat-reklamu-v-igry

🧠 Ещё больше инсайтов → в канале AFF.top

Discarding zero-volume keywords throws away your highest-converting terms
Tools report 0 or 'n/a' below a sampling floor, not true zero. We tracked 900 'zero-volume' long-tail queries for 90 days:
— Actually drew clicks ▇▇▇▇ 38%
— Median position when ranked ▇▇ 6.4
— Conversion rate vs head terms ▇▇▇▇▇ 2.3x higher
— Median real impressions/month ▇ 14
Long-tail means specific, and specific means closer to the buy. The tool just cannot sample it.
The mistake: filtering your export by volume greater than zero before you even look.
The fix: keep zero-volume terms that contain a buyer modifier (pricing, alternative, vs, for [use case]) and let the SERP confirm demand.
Benchmark of the week: zero-volume buyer terms convert 2.3x the rate of head terms in our cohort.

Embedding clustering groups 23% of terms n-gram methods scatter
Keyword grouping by shared words (n-grams) is fast but literal. Embedding-based clustering (grouping by meaning vectors) catches synonyms n-grams miss. We clustered 3,000 terms both ways.
— Embeddings merged 'laptop' + 'notebook computer' + 'portable PC': ▇▇▇▇▇ n-grams left these in 3 buckets
— N-grams over-split synonym families by 23%
— Embeddings can over-merge distinct intents (cost vs price vs free) without an intent guardrail
— N-gram grouping needs no model; embeddings need one API pass
So what: embed-cluster for topical structure, then layer a SERP or intent check to stop semantic-but-different terms from collapsing into one page.
Benchmark of the week: embedding clustering cuts the number of pages needed for a topic by ~20% versus exact-word grouping, before any quality loss.

CPC ranks commercial value, but mis-prioritizes 1 in 4 organic targets
Cost-per-click (the average advertiser bid) is a decent commercial-intent proxy, but it's an ad signal, not an organic one. We compared CPC-ranked vs conversion-ranked targets across 500 terms.
— High CPC, low organic conversion (ad-saturated SERP pushes organic below the fold): ▇▇▇▇ 26% mis-ranked
— Low CPC, strong organic conversion (advertisers ignore it, searchers still buy): ▇▇▇ underrated by CPC
— CPC reflects auction competition, not click value to a content page
— Still the best free proxy when you have no conversion data yet
So what: use CPC to sort cold lists before you have conversion data, then replace it with your own numbers the moment you have them. High-CPC terms often hand the click to four ads first.
Benchmark of the week: on high-CPC SERPs, organic result one sits below an average of 3.4 ad units, cutting its effective click share roughly in half.

Two terms with equal volume can be worth opposite amounts
A static volume number says nothing about direction. Across 800 matched pairs at the same volume:
— Rising term, 12-month slope ▇▇▇▇ +32%
— Declining term, same volume today ▇▇ -28%
— Value gap by month 12 of ranking ▇▇▇▇▇ 2.4x
— Plans that ranked terms identically on volume alone ▇▇▇ most of them
By the time your page ranks, the declining term has shed a third of its demand.
The mistake: ranking a roadmap on present volume with no slope attached.
The fix: multiply volume by trend direction; a rising 500/mo term outearns a sinking 800/mo term within a year.
Benchmark of the week: equal-volume terms diverged 2.4x in delivered traffic over 12 months.

На Anthropic, OpenAI, Google и SpaceXAI подали в суд из-за ИИ

В Калифорнии против ИИ-компаний подали антимонопольный иск: регулятору показалось подозрительным, что игроки синхронно призывают ограничить развитие нейросетей ради безопасности. Смысл спора в том, что инвестиции в ИИ уже обгоняют реальный прогресс, а бизнесу выгодны правила, которые защитят капитал. Вывод: быстрых прорывов ждать не стоит, лучше выжимать максимум из текущих инструментов.

➡️ Читайте на сайте: https://aff.top/blog/na-anthropic-openai-google-i-spacexai-podali-v-sud-iz-za-ii

🧠 Ещё больше инсайтов → в канале AFF.top

Google ads начал показывать расходы конкурентов

Google Ads запустил Peer Spending — инструмент, который сравнивает расходы аккаунта с рекламодателями из той же ниши без раскрытия чужих данных. Он показывает, тратите вы больше, меньше или примерно на уровне конкурентов на уровне кампаний и групп объявлений. Для арбитража это скорее ориентир по бенчмаркам, чем инструмент прямого усиления залива.

➡️ Читайте на сайте: https://aff.top/blog/google-ads-nachal-pokazyvat-raskhody-konkurentov

🧠 Ещё больше инсайтов → в канале AFF.top

Daily rank tracking adds noise, not signal, below position 10
Tracking frequency is a tradeoff: more data versus more noise. We logged the same 300 keywords daily and resampled weekly.
— Daily checks on page-2+ terms: ▇▇▇▇▇▇▇ swung 4-8 positions on personalization/location noise with no real change
— Weekly sampling smoothed that to a readable trend: ▇▇▇
— For page-1 terms near a money position, daily caught real drops worth a same-day fix
— Daily tracking multiplies credit cost 7x for the same conclusions on deep terms
So what: track money page-1 terms daily, everything else weekly. Daily data on deep rankings just buys you expensive noise and false alarms.
Benchmark of the week: ranking positions below 10 show roughly 5 positions of day-to-day jitter with zero underlying change, so daily granularity there is wasted spend.

Seed-based discovery surfaces 4x more terms than competitor-based
Two discovery tactics, same niche. We seeded one run with 5 head terms, the other with 3 competitor domains, and counted unique usable keywords.
— Seed expansion (matching terms): ▇▇▇▇▇▇▇▇ 3,400 raw, 22% usable after intent filter
— Competitor gap (their rankings you lack): ▇▇ 820 raw, 61% usable
— Overlap between the two methods: only 14%
— Competitor method skews toward proven, already-monetized terms
So what: seed expansion gives volume and white space; competitor gap gives validated, lower-risk targets. Run both. The 86% non-overlap is the whole point.
Benchmark of the week: competitor-gap keywords convert to traffic ~2.3x faster on average because someone already proved the SERP rewards that intent.

Минфин РФ планирует выпустить собственный стейблкоин

Власти РФ обсуждают запуск рублёвого стейблкоина: сейчас решают, как его обеспечить, какие операции разрешить и будет ли на него спрос. Основной кейс — международные переводы, а не использование физлицами. Если проект доведут до запуска, он может стать частью новой криптоинфраструктуры и альтернативой токенам, привязанным к дружественным валютам.

➡️ Читайте на сайте: https://aff.top/blog/minfin-rf-planiruet-vypustit-sobstvennyi-steiblkoin

🧠 Ещё больше инсайтов → в канале AFF.top