Incent and Offerwall Campaigns Bought as Rank Mechanics

Incent traffic is not user acquisition and we will not sell it to you as user acquisition. It is two things: an input to a store's ranking algorithm, and a cold start for a network model that has no data on your app yet. Across the bursts we have run, D7 retention on offerwall installs sits between 2 and 5 percent against 18 to 25 percent for the same app's non-incent paid traffic. That gap is the expected result rather than a failure, and any plan that depends on those users staying is already wrong. Android only, for rank: We run rank bursts on Google Play. We do not run them on the App Store. Apple's Review Guidelines treat artificial manipulation of chart position as grounds for removal, and the fact that a vendor will happily sell you iOS chart installs does not change where the consequence lands. If someone has quoted you a top ten position in an iOS category, they are selling you a risk that sits on your account and not on theirs. How a burst is scoped: The question is never how many installs. It is how many installs inside how many hours, in which country and category, measured against what the current occupants of that position are doing. We pull the category's install velocity for the preceding two weeks first and size against it. If the resulting number is not affordable, we say the burst will not move the position rather than running a smaller one and reporting the spend as an outcome. Bursts run over 24 to 72 hours. The position decays afterwards, and what holds it is organic pickup, which is a function of your store listing and not of the traffic we bought. Offerwall, for events rather than rank: The second use is cost per engagement. Tapjoy, Adjoe, Ayet, AdGate and Prodege will pay users to complete a defined event inside your app: reach a level, complete a profile, list a first item, finish onboarding. This is worth buying when a specific downstream event is the bottleneck in your own funnel and you want to see whether users who reach it behave differently, or when a bidder is starved of that event and needs volume to train. It is worth nothing as a revenue channel and we price it accordingly. Fraud control, which the offerwall does not do for you: Emulator clusters, device farms and reward loop abuse all live inside offerwall supply. Every burst runs through the MMP fraud suite with new device rate, IP concentration and click to install time as standing checks. We agree the rejection window in the insertion order before the burst rather than arguing about it after. A vendor who will not accept a 30 day rejection window has told you what is in the traffic. Not included: Incentivised ratings and reviews, in any form, on either store. Keyword search-then-install packages on iOS. Anything sold as keyword ranking installs. Bot traffic dressed as incent. We have turned down work over each of these and expect to again. Who this is not for: Apps trying to substitute a burst for a launch. Teams whose board reporting adds these installs to paid UA in one column. Anyone who needs the retention figure from this traffic to look acceptable, because it will not, and pretending otherwise is how the next quarter's forecast breaks. What you get: A pre-burst sizing memo with the category velocity data behind it, including the case for not running. The burst itself. Then a post-burst note separating rank movement, organic lift and incent installs into three lines that never get added together.