Incent and offerwall campaigns on a separate ledger

Incentivised traffic has three jobs worth paying for: a ranking burst timed to something that is genuinely happening, supply-side liquidity for a marketplace that needs listings before it needs buyers, and cost-per-engagement reactivation of users you already own. It has one job we refuse, which is making a blended cost per install look better than it is. Incent and non-incent never appear in the same line of a report here. How a burst is actually run: Store ranking responds to download velocity inside short windows and decays within days, so a burst is planned against a date that already exists in your calendar, a launch, a partnership or a broadcast, and never as a standing tactic. Before we book any volume, the store listing has to be finished: the first three lines, the screenshot order, the localised keyword set. A burst multiplies whatever conversion rate you already have. Sending traffic at an unfinished listing spends the budget on the multiplication of a bad number. Offerwall and cost per engagement: We work through the established offerwall demand sources and pay on events rather than installs wherever the partner supports it, because an install from a rewarded wall is close to worthless and a completed second-session action is not. For marketplaces this is the only version of incent we recommend without hesitation: paying for a listing created, a profile completed, a first message sent, on the supply side of a market where the cold-start problem is the entire problem. How it is reported: A separate tracker link and cohort tag in your MMP, set up before the first euro is spent, so that no future analyst can accidentally merge these users into a channel report. Every report shows day-one and day-seven retention of the incent cohort against organic, so the size of the discount is visible rather than argued about. If your board deck reports a single blended acquisition cost, we will ask you to change the deck before we start, and we have walked away when the answer was no. Not included: Keyword installs, bot-driven ranking and any vendor selling search-position guarantees. These sit squarely inside what both stores treat as manipulation, the exposure is your app's listing rather than our invoice, and we will end an engagement if that traffic is bought elsewhere while we are running the account. Also excluded: review generation of any kind, non-incent media buying, and creative production. Who this is not for: Donation and nonprofit apps, where an incentivised donor cohort is not a donor cohort and we say so before the brief is written. Subscription products whose economics depend on day-thirty conversion, where rewarded users convert far enough below baseline that the exercise flatters only the install chart. Anyone who wants incent to hit a blended CPI target set by a finance team that has not been told the traffic exists. What the work actually is: Choosing the window, sizing the volume against the position you need rather than the volume the partner wants to sell, deciding which event to pay on, and reading the cohort afterwards with enough discipline to say the burst did not hold. Most of the value here is in refusing the version of this that clients arrive asking for.