DAI Buying Across Megaphone, Acast and Spotify Ad Network

Between 8 and 20 percent of a dynamic insertion budget is usually spent twice. The same show sells through its own network, through the Spotify Ad Network and through an exchange, and no frequency cap crosses all three. The listener hears you four times in an afternoon and your effective reach is a fraction of the number on the plan. Where the leak is: Publisher-direct inventory reappearing inside an aggregated network. Platform-level frequency caps that dedupe within a platform and nowhere else. Impressions counted against a download window nobody in the room agreed on. IAB Tech Lab v2.2 counts a download when at least sixty seconds of content is requested by a unique device inside a twenty-four hour window, deduplicated on IP and user agent. We put that definition into the insertion order in writing, because half the discrepancies we chase turn out to be two parties counting different things rather than anyone delivering short. Platform by platform: Megaphone, bought direct with targeted impressions and streaming ad insertion, is where show-level control is best and where duplication with the Spotify Audience Network is most likely, so those two are never run blind against each other. Acast is bought through its own marketplace and programmatically through AdsWizz, and the two paths carry overlapping inventory at different prices. The Spotify Ad Network holds shows you may already be buying direct, which is worth checking before you treat it as incremental reach. Triton covers streaming radio and carries the postcode-level targeting that podcast inventory cannot offer. DSP-side, we buy audio through DV360 or Xandr deal IDs only. Open exchange audio without show-level reporting does not go on the plan, whatever the CPM says. Reconciliation: Weekly, against Podscribe or ArtsAI, and against the publisher's own numbers where the show is bought direct. A discrepancy under 5 percent is noted. Over 5 percent it is queried in writing with the platform. Over 10 percent, unresolved for a week, the flight stops and we move the budget. Frequency is capped at four exposures per listener per week on the device graph rather than per platform, which usually means buying fewer platforms than the plan first proposed. Pacing is checked against episode publication schedules, since a show that drops weekly cannot absorb a flat daily spend and will either underdeliver or bunch your impressions into three days. We will not run: Open exchange audio with no show-level transparency. Music streaming inventory reported inside a podcast line item, which flatters a CPM and mixes two entirely different listening states. Brand-safety keyword blocks applied to episode transcripts that strip out a third of the available inventory, followed by a manual buy of the same shows because someone senior likes them. Anything above four exposures per listener per week, which is where our own response curves stop improving and complaint volume starts. Bonus impressions accepted in place of a rate correction. Not for: Budgets under roughly twenty-five thousand euro per flight, where the reconciliation work costs more than the waste it recovers and you are better served buying two shows direct. Teams who need a live dashboard, since podcast delivery reports settle over days and a real-time number is a number that will change. Buyers who want reach maximised at any frequency; that plan is easy to build and we are not the ones to build it. What you get is a buy where every impression can be traced to a named show, a frequency position that survives contact with the second platform, and a weekly reconciliation you can put in front of finance.