Rewarded Placement Design for Apps That Are Not Games

A rewarded placement earns its build in a non-game app when about a tenth of daily actives will start one without being nagged into it. Below that, the format costs more in engineering, support tickets and store-review exposure than the revenue returns, and the honest advice is a better interstitial cadence instead. The first week of this engagement decides which of those two you are, from your own session data, before anything is designed. The design question: In games the reward is currency and the economy already exists. In a marketplace or a content app there is no economy, so the reward has to be carved out of something the product already sells or already gives away, which makes it a pricing decision rather than an ad decision. Give away what you charge for and you have taught your best users not to pay. Give away something worthless and nobody watches. The placements that hold up sit on a third thing: a temporary advantage that expires, costs nothing marginal to grant, and is visibly scarce. Placements that have held up: In a resale app, a listing raised to the top of a category for twenty-four hours in exchange for a rewarded view, priced so that a seller with more than four active listings finds the paid bump cheaper than watching four videos. In automotive classifieds, a full history summary on one vehicle, once a day, with the subscription still selling the unlimited version to the people who need it. In kids-and-family, a parent-gated grant of one extra episode, non-personalized demand only, with the ad never standing between a child and content they already had. Sizing the reward: Frequency caps come out of your session length distribution, not out of a round number somebody liked. The first session gets a grace period with no prompt at all. Subscribers see the offer only where it grants something outside their plan, and never in a form that reminds them a free user reaches the same place by watching a video. Every prompt string ships in English and French with the French at least as prominent, which Quebec's language law requires and which SDK-default dialogs quietly fail. What I will not design: Anything that interrupts content a child is already watching. Any prompt whose decline path is smaller, dimmer or slower than its accept path. Any reward that removes an obstacle you invented in order to sell its removal, the artificial cooldown or the wait timer added last quarter. And I will not ship a placement without a paired cannibalization read against the subscription cohort, because a placement that lifts ad revenue while cutting trial starts is a loss recorded as a win. Not for you if: You want the format live for a launch three weeks out. The measurement window alone runs longer than that. You are a game, where the patterns transfer badly and there are people who do that work better than I do. Or your product team has already fixed what the reward will be and wants an implementer: if the reward is wrong, implementing it well makes the outcome worse. What lands: A written placement specification with trigger conditions, caps, grace periods and decline behaviour for every surface. Copy in both languages, including the decline path. The instrumentation plan, covering rewarded starts per daily active, completion, repeat rate within a session, and the paired subscription metrics that decide whether the placement survives. Then a four-week read after launch with a recommendation to keep it, resize it or remove it. I have recommended removal often enough that you should treat it as a live outcome rather than a formality.