Native Traffic Arbitrage Built to Survive a Policy Review

Arbitrage is a spread measured per thousand sessions, and most of it dies of policy rather than of arithmetic. At a blended cost per click of 0.012 euro you pay about twelve euro for a thousand sessions; with 2.1 pageviews per session and a nine euro page RPM you gross about nineteen, and what survives discrepancy, clawed-back invalid traffic and net-60 terms is thinner than the first version of the model. We build properties that hold up when Google looks at them, and we do not cloak. The model comes before the site: Geo, vertical and ad stack decide feasibility before an article exists. We build the unit economics with your real payment terms inside it, because net-60 from an ad partner against a card charged daily on the traffic side is a working capital question before it is a marketing one. Spreads that look viable at fifteen percent margin are often not fundable at your cost of money, and week one is a cheaper place to discover that than month four. The ad stack ladder: AdSense is the honest answer below a certain volume and we say so instead of selling a build. Ezoic in the middle. Mediavine at fifty thousand sessions a month, Raptive at a hundred thousand pageviews. Ad Manager with Prebid and a handful of demand partners once somebody is running the operation daily, because header bidding without an owner is a slower AdSense with more invoices attached. Layout and speed: Ad slots reserve their height so the layout does not move under the reader's thumb, below-fold units lazy-load, and interaction latency stays under 200 milliseconds with the ad script running rather than in a lab build without it. Session depth is what pays here. A page that jumps while it loads costs the second pageview, and the second pageview is the margin. Policy, in writing: No cloaking, including the geo-redirect version that people describe as something else. No auto-refresh on AdSense inventory; on Ad Manager, refresh at thirty seconds of active tab time with a refresh key-value so buyers can price what they are getting. Nothing that would fail Google's scaled content abuse policy. The ANA's 2023 programmatic work put roughly fifteen percent of spend on made-for-advertising sites; the buy side now filters that inventory, so an MFA build has a shelf life measured in months and we will not sell one. Not included: Content at volume. We write twelve to twenty articles to set the standard and the templates, then your editors carry it. An agency offering you two hundred articles a month is offering you the exact thing the spam policy update was written about. Also excluded: SEO strategy, social distribution, and any statement about how long a given geo and vertical spread will last, because nobody honest knows. Who this is not for: Anyone who wants a machine that runs without an operator. This needs a person watching daily spend against daily RPM and acting on the gap. Not for a domain with a prior policy action, which we will not build on at any price. And not for anyone whose plan quietly depends on the traffic source and the ad network never noticing each other, which is a bet with a known ending.