MGID, Revcontent and AdNow Campaigns Run at Site Level

Two lists decide a widget campaign, and neither of them is the network: the publisher site IDs you allow to bid, and the page the click lands on. On a first MGID build we block between 200 and 500 site IDs inside the first fourteen days, before any headline test starts, because at that stage supply quality moves cost per acquisition further than creative does. We buy MGID, Revcontent and AdNow, in Czech and English, and we decline the engagement when the landing page has to be an existing product page. How the buy is structured: Campaigns launch split by device and by operating system, because one site ID behaves like two different publishers on Android and on desktop, and a blended cost per click hides that. Opening bids are set high enough to buy data quickly, then come down per site once each has enough clicks to be judged on something other than luck. We reach for per-site bid coefficients before the pause button: a source that loses money at nine cents is often steady at four, and a paused source is supply you rarely get back at the same price. What we measure: The click id macro goes into Voluum or RedTrack, and a server-side postback returns the conversion to the network so its optimiser sees what your CRM sees. Browser pixels alone under-report and teach the network the wrong lesson for a month before anyone notices. Three numbers are reconciled every week: network clicks, tracker clicks, sessions in your own analytics. A gap above roughly fifteen percent between tracker and analytics is not rounding. It is a redirect losing people, or traffic that was never a person. The article: The widget rents attention on somebody else's page. The forty seconds after the click is where the campaign is decided, so we write the landing article ourselves, one page per angle, in the language of the market. It carries the argument an editor would make, and it says who paid for it. Clients who want the click sent to a category page are better served elsewhere; we have watched that arrangement fail often enough to stop accepting it. What we will not do: Chumbox creative: fake close buttons, before-and-after bodies, borrowed publisher logos, arrows drawn on faces. Promotion of prescription medicines to the Czech general public, which Act No. 40/1995 Coll. prohibits outright. Run-of-network buying in an account where we do not control the blocklist. Accounts opened in our name rather than yours. Not included: Taboola and Outbrain, which are a different buy with different minimums and where our blocklists transfer badly. Programmatic display, paid search, SEO and video production. Engineering on your own stack beyond the article itself. Media budget, which is paid from your card directly to the networks. Who this is not for: Budgets below about eight thousand euro a month in media, where the networks never leave the learning phase and you pay the tuition without getting the lesson. Affiliates who need a funnel we cannot put our name on. Brands where each article revision waits three weeks on a legal committee, because the buy moves faster than that and the cost of the delay lands on you rather than on us.