Verification Stack Setup: IAS, DoubleVerify and Life After Moat

Moat is not an option. Oracle shut its advertising business, Moat measurement included, on 30 September 2024, so any verification project that names Moat today is a migration project. The work is not installing tags. It is rebuilding segment logic and, harder, rebuilding a historical baseline in IAS or DoubleVerify that your trend reporting can sit on top of without a step change nobody can explain to finance. Pre-bid and post-bid are different purchases: Pre-bid avoidance segments carry a CPM fee and shrink the auction you are bidding into. Post-bid tags measure and prevent nothing. Most accounts pay for both and have never compared the count of impressions the pre-bid segment actually blocked against the fee paid to block them. We produce that comparison before setting a single threshold, and in some accounts the conclusion is pre-bid switched off on the channels where it was buying nothing. Portuguese is where these tools are weakest: Two failures repeat. The first is an English keyword blocklist machine-translated into Portuguese, which blocks ordinary football reporting and misses the pt-BR terms that would have mattered. The second is a generic gambling category applied to a licensed betting advertiser, which blocks precisely the inventory that converts. Sports content classified against violence terms is a close third. We rebuild the list in Portuguese for the specific advertiser and test it against a sample of your own recent supply rather than against a vendor demonstration. Which standard you are actually holding: The MRC counts a display impression viewable at fifty percent of pixels for one second, and video at fifty percent for two. GroupM's definition is stricter and produces a lower number from identical inventory. CTV has no equivalent viewability standard worth quoting, so what you police there is invalid traffic, app labelling and completion. GARM closed in August 2024 and its category framework survives inside vendor taxonomies, which means GARM-aligned now means whatever your vendor decided it means. We write the floors into the insertion order in words rather than by reference to a body that no longer meets. Reconciliation: DSP-reported viewability and vendor-measured viewability disagree, routinely by a few percentage points and occasionally by more when a wrapper is timing out. One of those is the contract number. We decide which before launch and put the tolerance in writing, because the alternative is a quarterly argument in which both parties are quoting a true number at each other. On attention metrics: We will implement an attention measure and we will not let it replace viewability in the contract until it has been calibrated against your own outcome data across a full quarter. Buying against an uncalibrated attention score means buying against a vendor's model of your business, which is a strange thing to pay for twice. What is not included: We hold no vendor contract in our name and take no rebate, referral fee or reseller margin from IAS or DoubleVerify. That clause is in our agreement rather than in a conversation. No fraud litigation support or make-good negotiation with exchanges. No MRC accreditation work. No creative or media buying. Who this is not for: Buyers who want a hundred percent viewability guarantee, which several vendors will sell you and which means the inventory was selected for measurability rather than for people. Advertisers who want blocking set so hard the campaign cannot spend and then want the delivery shortfall explained. Anyone who needs the verification report to say something the log files do not.