Push Network Buying on PropellerAds, RichAds and Zeropark

A three-network push test at nine thousand dollars of media resolves in twelve to fourteen days, and what it returns is a source list rather than a verdict. Across a Southeast Asian geo set, spend touches several thousand publisher zones and the ones that pay settle into a group of forty to sixty. Everything after that fortnight is holding that group while it decays, which it does. Why three networks and not one: PropellerAds, RichAds and Zeropark are not three doors onto the same inventory. PropellerAds carries the widest publisher base and a CPA Goal bidder worth handing control to only after a manual phase has shown the offer converts at all. RichAds sells its subscriber base split by how recently a device accepted the prompt, and the freshest slice behaves like a separate product with its own economics. Zeropark is narrower and answers to target-level and source-level rules more than to bid pressure. Same offer, same creative set, three demand paths, and by week three the spend split tells you which network is actually yours. Tracking is the condition of entry: Voluum, Binom or Keitaro, with server-to-server postbacks firing on the event you are paying for: booking confirmed, mandate signed, donation captured. Not landing page arrival. We decline engagements that intend to judge push inside GA4 last-click, because the click identifier does not survive the redirect chain, the channel then reads as direct traffic, and somebody senior concludes it is worthless. If your stack cannot pass a postback, that is a project, and it is a different one from this. Subscriber age, priced honestly: Devices within a week of opting in cost more per click and convert at a multiple of subscribers past ninety days. For anything with deliberation in it, a fare, a tariff switch, a recurring giving mandate, the fresh tier wins on cost per acquisition at roughly double the click price. We show that arithmetic per geo every week rather than asserting it, and where it stops holding we move budget to the older tiers and say why. How zones get cut: Zones are judged on cost per conversion against target, and never before they have spent three target CPAs. Below that the number is noise dressed as a decision. We do not prune on click-through rate. On push a high rate is as often a thumb landing on a notification being dismissed as it is interest, and the zones with the best rates are frequently the worst payers. What we refuse: Creative that imitates a system message, a chat badge, a parcel notice, an antivirus warning or a bank alert. It wins split tests. It is also why the format has the reputation it has, and a company with a regulator, a board or a donor file pays that bill later at a price nobody quotes upfront. We do not run nutra, dating, crypto giveaway or sweepstakes offers, and we do not arbitrage adult inventory into a mainstream funnel. Who should not hire us for this: Affiliates whose margin depends on the creative described above. Anyone wanting a guaranteed cost per acquisition, since push supply moves week to week and a guarantee at this price is either padded or untrue. Teams who want logins handed over and no interpretation attached. What arrives, weekly: Spend, conversions and cost per conversion by network, geo, zone group and creative. The whitelist and blacklist as files you import yourself and keep. A note saying what changed and why it changed. Buying is done from Singapore, in English or Mandarin, by the person who wrote this.