В Telegram Ads добавили Banner in Bot

В Telegram Ads появился формат Banner in Bot для показа рекламы внутри ботов с аудиторией от 1000 человек. Инструмент таргетируется не на площадки, а на пользователей — по интересам, номерам телефонов и подпискам. Пока доступны только текстовые креативы по ставкам Target Users. Новинка позволяет напрямую через стандартный кабинет охватывать целевую аудиторию прямо в их диалогах с ботами.

➡️ Читайте на сайте: https://aff.top/blog/v-telegram-ads-dobavili-banner-in-bot

🧠 Ещё больше инсайтов → в канале AFF.top

Anthropic готовит к запуску Claude Money

Anthropic анонсировала Claude Money — ИИ-сервис для управления личными финансами и автоматизации платежей. Инструмент позволяет подключать банковские карты для анализа расходов, планирования бюджета и проведения транзакций. Это переход от консультационных моделей к полноценным финансовым агентам. Внедрение технологии позволит пользователям делегировать нейросети рутинные задачи, включая оплату токенов и подписок на необходимые рабочие сервисы.

➡️ Читайте на сайте: https://aff.top/blog/anthropic-gotovit-k-zapusku-claude-money

🧠 Ещё больше инсайтов → в канале AFF.top

The survey question that flipped the story

A remote-work software brand commissioned a 2,000-person survey and almost buried the one finding that mattered. The setup: the deck was full of safe, expected stats, the kind reporters delete. The angle emerged from a single counterintuitive line in the data: 41% of managers admitted they trusted in-office staff more, even when remote staff outperformed them.

That tension was the whole pitch. We dropped the comfortable numbers entirely and led with the contradiction: 'Managers say remote workers are more productive, then promote the people they can see.' Conflict travels.

The result: 47 placements across business and HR press in three weeks, including Fast Company, Harvard Business Review's site, and a BBC Worklife feature, 38 with links. The survey landing page pulled 19,000 sessions. An HBR editor noted, 'The bias number is the only one I hadn't seen ten times this year.'

We later A/B tested headlines and the contradiction angle outperformed the productivity stat by nearly 4 to 1 in reply rate.

Lesson: Surveys don't fail for lack of data. They fail when you lead with the answer everyone expected.

We inherited a media list of four hundred names, and it was worse than having none

A new client handed us a contact list their last agency built — four hundred reporters, neatly tagged. The alternative was building our own from scratch, slowly, by reading. We tested both on the same campaign to settle the argument.

The inherited list was seductive: huge, instant, ready to blast. We sent a hundred pitches off it and got a 2% reply rate and one angry 'how did you get my email.' Half the contacts had changed beats or outlets. The list was scale without relevance — a spray.

The built list was twenty-five names, assembled over a week by actually reading recent articles and noting who covered our exact topic. We pitched personally, referencing each reporter's recent work. The reply rate was 28%, and three of them remembered us for later stories. Twenty-five right names beat four hundred wrong ones every time.

"You'd clearly read my last piece," one replied. "That's why you're getting an answer."

Lesson: a big inherited or bought list is scale with no relevance and rotting data — it converts like spam because it is spam. A small list you build by reading is slow but every name is a genuine fit, which is the only thing that moves reply rates. Never confuse the size of a list with the quality of a campaign. Twenty-five beats four hundred.

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

The breaking story gave us an hour, and we had to choose speed or substance

A regulation dropped at 9am that touched our client's industry directly. Newsjacking forces a brutal pick: fire off a fast reaction now, or take three hours to build a substantive one. We'd done both on different stories, so we knew the curves.

The fast reaction — a sharp 80-word comment out within forty minutes — caught the first wave of reporters still writing their initial pieces. Two used it. The catch: a quick reaction is, by nature, shallow, so it earned a single buried quote each, the kind any competitor could've supplied.

The deep reaction, on a previous story, took us three hours to assemble with original analysis and a small data pull. We missed the first news wave entirely. But we caught the second — the 'what does this mean' explainer pieces that come a day later, which need exactly the substance a fast take can't carry. Those earned standalone features, not buried quotes.

"Everyone had a reaction by noon," one reporter said. "By the next morning I needed someone who'd actually thought about it."

Lesson: the fast reaction catches the first wave and wins thin, interchangeable quotes — speed is the only edge, so be first or don't bother. The deep reaction misses wave one and owns wave two, where analysis becomes the whole story. Know which wave you're staffing before you start typing.

The benchmark that made rivals quote us

A call-center software vendor was invisible in a category dominated by two giants. The setup was a David with no megaphone. The angle was to become the neutral scorekeeper: build an independent benchmark of average customer-wait times across the whole industry, including the giants, with the methodology fully open.

The pitch positioned us as referee, not player: 'We measured average hold times across 50 brands. The worst keeps customers waiting 11 minutes; the best, 40 seconds. Full data attached.'

The result: 35 placements in customer-experience and business press, 28 with links, and the surprise twist, the two market leaders cited our benchmark in their own marketing, linking back to legitimize their good scores. Being the source everyone trusts beat being the loudest. The benchmark drew 13,000 sessions and 19 inbound link requests.

A CX trade editor said, 'You're the only vendor whose numbers I'd run without checking, because you included your own weak spots.'

Lesson: Rank yourself honestly alongside rivals and you become the citation. Cook the benchmark and you become a press release nobody links.

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