Rewarded video campaigns measured on day-7 retention

Rewarded video completes above 90 percent in almost every placement, because the viewer agreed to the trade before the first frame played. That figure describes the format, not your ad, and it does not appear in anything I send you. These campaigns are bought, cut and reported against one of two events: a day-7 retained user, or a first completed action inside your own system — a load accepted, a booking paid for, a shift claimed. If nothing in your product gives someone a reason to come back inside a week, rewarded is the wrong format and I will say so before you write a brief. Where the format earns its place: Driver and warehouse recruiting for carriers, 3PLs and last-mile operators. The people you need are in a mobile game at nine in the evening and not on a job board, and rewarded is the only placement where they will watch thirty seconds of a recruiting pitch on purpose. Travel apps inside a live booking window, where the return visit is already going to happen and the only question is whose app it happens in. Re-activation against a lapsed base you can name. Reward design comes before any media plan: The reward decides who shows up. Something worth farming brings people who farm. Something too small gets dismissed at the option screen and you pay for the impression regardless. I ask for the app economy documentation or the currency ledger before writing a plan, then place the unit at a resource wall or a fail state. A rewarded slot on first launch sells to people who will delete the app that night, and it will look like your strongest placement for about four days. Supply, and what gets cut: AppLovin, Unity LevelPlay, Liftoff, Mintegral and AdMob, bought in bidding where bidding exists rather than through a legacy waterfall. Cuts happen at sub-publisher and placement level, read off the raw report; cutting at network level throws away the good half with the bad. Bundle IDs get checked against app-ads.txt, and anything that will not resolve to a real listing comes off the plan before flight. Measurement, set up before the first dollar: On iOS, the SKAdNetwork and AdAttributionKit conversion value schema is written against your retention event and locked before launch, with the postback window chosen to match how long your product takes to matter. A 24-hour window on a product whose first real action lands on day three tells you nothing at all. On Android, Play Install Referrer joined to a server-side event you already own. Where your geographic footprint allows one, a matched-market hold-out, because attributed installs and caused installs are different numbers and only one of them is yours. Not included: Playable ad production, app store listing work, and in-app event instrumentation. If your events are not firing, that is engineering work and it comes first. I do not buy offerwall or cash-reward supply: that audience is working rather than watching, and the result reaches your dashboard labelled as installs. I do not accept installs as the reporting currency, and I do not send a weekly report that opens with impressions. Who this is not for: Developer tools. I work in that market and have never seen rewarded inventory produce a paying team, with one exception — engineering recruitment, where the audience overlap is genuine. It is also wrong for anyone without an app or a logged-in web product, and wrong for a brand that wants its logo held on screen, because the viewer is counting down to the reward and will say so in the reviews.