> ## Content Index > Fetch the complete content index at: https://bielawski.pro/llms.txt > Use this file to discover other available public pages before exploring further. # Press review 09.11.2026: Cloudflare charges AI companies for page access, and Tripadvisor counts the cost of losing Google traffic - URL: https://bielawski.pro/en/press-review-09-11-2026-cloudflare-charges-ai-companies-for-page-access-and-tripadvisor-counts-the-cost-of-losing-google-traffic/ - Published: 2026-11-09T06:08:42.000Z - Updated: 2026-11-09T06:08:42.000Z - Description: Cloudflare has booked its first revenue from charging AI companies, and OpenAI already pays for signals about page changes. Tripadvisor writes off $369.5 million and admits SEO will keep weighing on growth in 2027. - Author: Bartosz Bielawski - Tags: Prasówka, #en ## Trending topics Topics that clearly more independent sources have covered in the last two weeks than before: - [AI agents have stopped being chatbots - now they work in your company, your browser and your reports](https://bielawski.pro/en/ai-agents-have-stopped-being-chatbots-now-they-work-in-your-company-your-browser-and-your-reports/) ## SEO and AI search ### Tripadvisor writes off $369 million because the drop in search traffic is no longer temporary The travel giant booked a $369.5 million goodwill impairment in the third quarter and named the reason itself: search traffic. Revenue in its hotels segment fell 23%, to $151.6 million. The more interesting part is the cost structure. Marketing consumed 48.8% of revenue, even though spending stayed flat in dollars. The campaign did not grow - the revenue base shrank. AI traffic is still under 1% of visits, but it converts several times better than average. The CFO said SEO will remain a headwind to growth in 2027 as well. What to do: work out what share of your client's revenue depends on SEO and what it would cost to replace that traffic with paid. Only that number tells you whether a decline is a problem or just seasonal. **Sources:** - Search Engine Journal, 2026-11-08: [Tripadvisor Says SEO Will Still Weigh On Growth In 2027](https://www.searchenginejournal.com/tripadvisor-says-seo-will-still-weigh-on-growth-in-2027/596257/?ref=bielawski.pro) ### Cloudflare makes AI companies pay to access your site and calls it the new version of SEO Cloudflare reported its first revenue from products that charge AI companies for access to your site. Revenue in the third quarter reached $782.5 million, up 39% year over year. This is not a promise for later. OpenAI already pays Cloudflare for signals about when a page has changed, so its crawler does not fetch the page needlessly. CEO Matthew Prince expects other AI companies to follow. For a site owner, this is a business decision, not a technical one. You can block selected bots (AI Crawl Control is available on every plan), charge per visit, or let agents in and count on traffic and citations. What to do: decide whether you want agent traffic or want to block it, and check whether your hosting has already made that decision for you with a default setting. **Sources:** - Search Engine Journal, 2026-11-08: [Cloudflare CEO Likens AI Agent Optimization To A New Version Of SEO](https://www.searchenginejournal.com/cloudflare-ai-agent-optimization-new-seo/596248/?ref=bielawski.pro) ### A business with 205 reviews loses to one with 40 in Google Maps - because of a single label *New signal: one source so far.* The primary category in your Google Business Profile decides which queries Google will even consider you for. A roofer listed as a "general contractor" loses to a competitor listed as a "roofing contractor" for "roofer near me", even though he has five times as many reviews. This is not a detail that reviews can make up for. The label is the first thing Google reads to decide what you do, and in Maps it is often what appears under the business name. What to do: open Maps on your phone in a private window, search for your main service and your town, and compare the labels under the top three results. Then check whether your category completes the sentence "this business is a..." rather than "this business has a...". **Sources:** - Missoula SEO Geek, 2026-11-07: [200 Reviews and Still Losing on Google Maps](https://www.youtube.com/watch?v=B1HXu1dTI%5FU&t=0s&ref=bielawski.pro) - Missoula SEO Geek, 2026-11-08: [Check Your Primary Category Tonight](https://www.youtube.com/watch?v=K9dZlSAo61k&t=0s&ref=bielawski.pro) ## Marketing and content ### The call count in your Google profile measures taps, not calls - check how many you are losing Google counts every tap on the "call" button, not the conversations that actually reached the business. In the dashboard it looks like success, even when half of those people hit a busy line or an answering machine. The difference matters, because every tap is someone who wanted to call. If they did not get through to you, they most likely called a competitor, while all you see is a nice number. What to do: call the number in your profile and check whether a person answers during the hours you have listed. Then compare the tap count with real enquiries in your CRM or call log. **Sources:** - Missoula SEO Geek, 2026-11-08: [Google's Calls Number Counts Taps, Not Calls](https://www.youtube.com/watch?v=yQ0vbFnw4s0&t=0s&ref=bielawski.pro) ### 32 Facebook ad formats: a list to pick three from, not one to use in full A Polish e-commerce channel has compiled 32 creative formats that work in Facebook Ads in 2026\. Among them: "problem-solution", "myth versus fact", "customer review", "don't buy this if" and "founder calls the customer". The value is not in the number. A format is just a way of telling one thing: if the creative has no clear angle, changing the template will not help. What to do: pick three formats that match the buying stage your product sits in, and test them on the same budget instead of rotating all of them at once. **Sources:** - Borysewicz Ecommerce, 2026-11-08: [Ta lista jest wartościowa nie dlatego, żebyś użył wszystkich 32 formatów](https://www.youtube.com/watch?v=-pUCmEnkUMk&t=0s&ref=bielawski.pro) ## AI and agents at work ### 17 AI agents cut a project from 9 months to 2.5 - but only after the process changed Levi Strauss showed what deploying agents in everyday work looks like. Seventeen specialised agents handle requirements, testing, accessibility and security, while humans keep two decisions: approving the specification and the final merge. The results are measurable: throughput rose 51%, and the share of AI-assisted changes grew from 26% to 69%. The strongest resistance came from the best people on the team, who had to unlearn what had made them successful. The team's conclusion is blunt: an agent is a markdown file in a repository, changed through a normal pull request. A new tool inside an old process delivers about as much as an electric motor in an old factory. What to do: before you deploy an agent, write the process down and mark where a human makes decisions and where a human only executes. **Sources:** - GitHub, 2026-11-08: [17 agents, one SDLC: How Levi Strauss reimagined development on GitHub](https://www.youtube.com/watch?v=-GpxKKgoWWg&t=1537s&ref=bielawski.pro) ### Agents already create half of the pages in Notion, and AI usage has grown ninefold Notion announced a new product chapter and shared two numbers along the way: AI usage inside the app has grown ninefold, and roughly half of all pages are created by agents. The direction is clear - Notion no longer wants to be the place where work happens, but the layer that agents draw company knowledge from. For marketing teams, this is a signal that documentation is no longer just an archive for people. Increasingly, a model reads it, so the structure, naming and freshness of notes now affect the quality of an agent's work. What to do: tidy up one knowledge base your team uses with an agent in mind - clear titles, one source of truth per topic and an update date inside each note. **Sources:** - Tool Finder, 2026-11-08: [Notion's Chapter 4?](https://www.youtube.com/watch?v=npnm1LzBEa4&t=0s&ref=bielawski.pro) ### Klarna saves $40 million with AI, but comparisons like this confuse two things The same numbers keep coming up in the AI debate: Klarna replaced 700 customer service staff and saves $40 million a year, JP Morgan saved 350,000 lawyer hours, and PayPal cut fraud losses by $700 million. The problem is that such comparisons pit a fresh deployment against a process refined over years. A fair comparison needs the same amount of time and the same number of iterations on both sides. The second thing that does not work: advertising that you use AI. Clients care about faster, cheaper and less risky, and in B2B the missing piece is proof that someone accountable stands behind the polished answer. What to do: pick one part of the process, automate it all the way through, then measure it with the same yardstick you use for people. **Sources:** - MoreMozi, 2026-11-09: ["Are You Comparing New AI With Years of Human Process Training?"](https://www.youtube.com/watch?v=kAT-o%5FLR6Q8&t=385s&ref=bielawski.pro) ## Business and sales ### Divide new customers by churn and you will see where your business stops growing It is simple division: new customers per month divided by monthly churn gives you the ceiling a business can grow to. With five new customers and 10% churn, the ceiling is 50 customers, no matter how good your product is. The same maths works for money. Average revenue per customer divided by churn gives you lifetime value, and new monthly revenue divided by churn gives you the ceiling on monthly revenue. In practice, companies reach 80-90% of that maximum. What to do: calculate your ceiling this week. If you are close to it, more acquisition budget will not help until churn falls or referrals rise. **Sources:** - MoreMozi, 2026-11-08: ["Do Current Inflow and Churn Support Your Growth Target?"](https://www.youtube.com/watch?v=mGLB8LErfxA&t=209s&ref=bielawski.pro) ### A premium offer nobody buys still lifts sales of your cheaper services A high-priced, unscalable offer at the top of your price list changes the reference point for everything else. Even if nobody buys it, clients compare the cheaper package with that price rather than with your competitors. The author's maths is simple: when 10% of clients buy a version that costs ten times as much, company revenue doubles, and almost all of that extra revenue is margin. There is one condition - you have to say the price out loud and let the client think about it, instead of offering the cheaper version right away. What to do: add one premium service to your price list at ten times the base price and watch how the conversation about the cheaper package changes. **Sources:** - MoreMozi, 2026-11-08: [How to Design and Present a Higher-Value Service](https://www.youtube.com/watch?v=54kBP6tQFS4&t=150s&ref=bielawski.pro) ### Changing just the headline tripled a page's conversion rate - the rest mattered less Over three weeks of funnel tests, more than 100 page elements were tested. The biggest swings came from the headline and the opening sentences, not from anything below the first screen. In one case, the headline alone tripled conversions; in another, removing the headline from a scheduling page increased them fivefold. The author's conclusion is uncomfortable for anyone who likes polishing the bottom of the page: 90% of attention goes to the first screen. What to do: in your next test, change only the headline and the first line of copy, not the whole layout. One element, one hypothesis, one result. **Sources:** - MoreMozi, 2026-11-09: ["Are You Testing the First Screen or Polishing Below the Fold?"](https://www.youtube.com/watch?v=l95J0GY2Cw4&t=496s&ref=bielawski.pro)