Verification stack rebuild after the Moat shutdown

Oracle shut its advertising business in September 2024 and Moat closed with it, so a Moat setup is now a migration, and the first thing to plan for is the discontinuity. Moving the same inventory from one vendor's measurement to another's typically moves reported viewable rate by three to eight points. Nothing about the media changed. Your trend line did. If that number reaches a board, we agree the changeover date and the restated baseline before anything is tagged. What I build: Pre-bid comes first, because it is the only part that saves money rather than describing it. IAS or DoubleVerify segment keys go onto line items in The Trade Desk and DV360 with an explicit split between avoidance and targeting, and with suitability tiers written down as a document your agency can be held to. Since GARM closed in August 2024, that mapping sits against the framework the 4A's and ANA now maintain rather than a vendor's own tiering, which drifts between releases and is not something you can hold anybody to. In-app is where most stacks are quietly broken. Open Measurement SDK has to be present in the app, initialised by the app, and correctly passed through whichever mediation layer is already running — Google Mobile Ads, AppLovin MAX, ironSource. The AdVerifications node in VAST 4.x has to survive the wrapper chain, which it frequently does not once a second wrapper is added by an intermediary. I test that end to end with a controlled buy against your own inventory before calling the integration done, because a vendor dashboard will report an acceptable-looking measurement rate while an entire mediation path goes unmeasured and silently drops out of the denominator. Post-bid tags come last, wrapped in a fixed order and documented. Discrepancy between DSP-counted and vendor-counted impressions is normal; the question is the band. I set the tolerance, write down what sits inside it, and define which SSP conversations get escalated when a path falls outside. The standard you pick: The MRC counts display as fifty percent of pixels for one second and video as fifty percent for two. Some advertisers hold their buys to one hundred percent for one second. That is a commercial decision with a CPM attached, not a platform toggle, and it belongs in the insertion order. I will tell you what each choice costs against your current supply, then you choose it. Not included: I do not run the verification account as an ongoing service. After handover your team or agency operates it. I would rather build something you can hold a vendor to than become the person who logs in. I do not write keyword blocklists. Long keyword lists mostly defund news, cut reach, and change no outcome I have been able to measure. I do not certify results against a vendor's own definition of success. Wrong fit: Under roughly 150 thousand US dollars a quarter of programmatic spend, the vendor licence plus this work does not return. You are better served by a tighter supply path and no verification layer at all, and I will say so on the first call rather than sell you the smaller package. If the requirement is that measured viewability rises this quarter, hire someone else. Measurement built so a number moves is the reason the number stopped meaning anything. Languages: English and Mandarin. Singapore hours, with US Eastern mornings when the account needs the platform calls.