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

Paid rank tracker vs GSC average position. They lie differently.
Client screams 'we dropped to #8!' Rank tracker says so. GSC says avg position 4.2. Who's right? Both. Neither.
— Rank tracker: one location, one device, one moment. Clean for spotting a real algo move on a single term.
— GSC avg position: blended across every query, geo, and device that page ranks for. Truer picture of actual visibility, useless for one keyword.
My rule: track 10-20 money terms in a tracker for sharp daily signal. Judge OVERALL page health off GSC clicks + impressions, never avg position alone.
Stop reporting avg position as a KPI. It's an average of averages. Use clicks.

Killed 40% of a site's pages. Sitewide traffic went UP 28%.

Sounds insane. Site had 900 pages, half were thin tag archives and 300-word zombies pulling 0 clicks for 12 months straight.

My move:
— Pulled GSC, flagged every URL with under 3 clicks in 6 months
— 301'd the salvageable ones into related hubs
— noindex'd the rest (tag pages, empty category pages)

Crawl budget collapsed onto the pages that mattered. 5 weeks later the money pages climbed 1-3 spots each. Sitewide clicks +28%.

Dead weight drags your whole site. Go audit your zero-click pages and cut.

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

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

Topic clusters vs one-page-per-keyword. The second still works — sometimes.
Clusters are gospel now, so everyone forgot one-keyword-one-page can still win. Pick by competition depth.
— Low-competition, high-intent long tail ('how to cancel [tool] subscription'): a single tight 800-word page nails it. No cluster needed, the intent is atomic.
— Competitive head term: you CAN'T win it with one page. You need the pillar + 10 spokes proving topical depth. Google won't trust a lone page on 'project management software'.
The tell: how many ranking competitors have a whole cluster? If the SERP is single pages, match it. If it's clusters, you need one.
Look at your next target's SERP. Count the clusters.

Added a comparison table to a buyer post. Conversions up 41%, rank up too.

Review-style post ranking position 6, low affiliate clicks. Wall of text, no scan-ability.

I dropped in one HTML comparison table near the top:
— 5 products, 6 columns (price, best-for, rating, key feature)
— Each row linked out to the affiliate offer
— Sticky enough that people stopped scrolling to read it

Affiliate click-through up 41%. Bonus: dwell time rose, and the post climbed 6 to 3 over the next month.

Buyers want to compare, not read essays. Give them the table up top. Go add one to your best money post.

Publishing in bursts is wrecking your crawl rhythm
Did this for a year: dump 8 posts the last week of the month, then radio silence. Googlebot learned to come around once a month and miss the fresh stuff for weeks.

Switched a client to boring, predictable cadence — 2 posts/week, same days:
— Crawl frequency tightened up within ~3 weeks (checked the logs)
— New posts got indexed in days, not weeks
— Output stayed identical, just spread out

Google rewards a heartbeat, not a flatline with spikes. Set a schedule and hold the line.

AI to draft vs AI to edit. I flipped which one and quality jumped.
For a year I had AI draft and me edit. Output was fine. Forgettable. Indexed slow.
Flipped it: I draft the bones rough and human, AI does targeted EDIT passes.
— Pass 1: 'flag every claim with no source or example.' Forces my experience back in.
— Pass 2: 'tighten, kill throat-clearing, no new info.'
— Pass 3: 'list questions a reader still has.' That becomes my FAQ.
AI as drafter averages you into the SERP middle. AI as editor sharpens what's already yours. The human-first draft is what Google's helpful-content stuff is sniffing for.
Flip the order on your next post. Feel the difference.