“Two-thirds of deals are configured with the desired supply, yet deliver no or very little revenue.”
That is Magnite’s VP of Product Management, on the record, in IAB Tech Lab’s Deals API announcement of 4 December 2025. Not an analyst estimate. An exchange executive describing the state of the deal layer in a standards release.
Hold it against the word guaranteed, and the appeal of the whole category gets harder to state. The emotional pitch of a guaranteed deal is safety: a signed, correctly configured commitment is money in the bank. Two thirds of configured deals are not.
That is not a sales problem. It is a pricing and forecasting problem, and the spec documents both sides of it more honestly than the marketing does.
What guaranteed actually means in the protocol
OpenRTB 2.6 defines it in one line. On the Deal object: “guar | integer, default 0 | Indicates that the deal is of type guaranteed and the bidder must bid on the deal, where 0 = not a guaranteed deal, 1 = guaranteed deal.”
The auction type field is where the money is decided. At deal level, at can take the value 3, meaning “the value passed in bidfloor is the agreed upon deal price”. No discovery. The number you negotiated is the number you get.
Google’s Authorized Buyers documentation says the same thing from the buying side, and it is worth reading twice: “Fixed_cpm_micros represents the deal terms CPM that was agreed by both parties. At this time, we ignore the bid price (in the response) in the auction and always clear the bid at the deal terms CPM.”
So on a guaranteed impression the buyer’s willingness to pay is discarded. If that impression would have cleared at 3 times your negotiated rate, you will never know, because nothing in the path records the counterfactual.
IAB Tech Lab’s own implementation guidance warns exchanges against treating deals as a price contest at all: “Deal ID bids may not be subject to a classic price auction. There may be an expectation that the buyer and seller want prioritization to achieve a larger objective: complete delivery of the Deal.”
That is the trade, stated in the standard. You exchange price discovery for delivery certainty.
The share of the market that takes that trade is smaller than you think, and it is shrinking
FreeWheel’s Video Marketplace Report for the second half of 2025 puts guaranteed at 18 percent of US programmatic premium video ad views, against 82 percent non-guaranteed. In Europe the split inverts: 62 percent guaranteed, 38 percent not.
And the US number moved the wrong way for the guaranteed thesis, from 20 percent in the first half of 2025 to 18 percent in the second. Different halves are not seasonally like-for-like, and this is premium video rather than mobile in-app, so read it as direction rather than as a benchmark.
Which brings up a genuine hole in the public record: there is no published split of guaranteed against auction-cleared spend for mobile app inventory. FreeWheel’s data is video and CTV. PubMatic reports that mobile app revenue grew over 40 percent year over year and reached about 25 percent of total revenue in Q2 2026, but publishes no deal-type breakdown. Nobody does. Anyone quoting a mobile in-app guaranteed share is extrapolating.
The asymmetries that sit inside the contract
Three of them, all documented by the platforms themselves.
The buyer’s side has an obligation too, and it is rarely monitored by the seller. Adobe lists the buyer’s responsibility as “Executing a 95%-100% bid/response rate”, and OpenRTB’s must-bid flag exists precisely because buyers do not always honour it.
Four things worth doing before signing the next one
Where the exchange relationship matters
Every decision above depends on knowing your own clearing price distribution well enough to price certainty against it. That is a data question before it is a commercial one, and it gets harder the more intermediaries sit between the impression and the bid.
For example, at AdSpin the exchange runs its own bidder on OpenRTB across owned and operated inventory plus direct publisher supply, so clearing prices are observed at the source rather than reconstructed from a partner’s monthly report. That does not make guaranteed deals better or worse. It makes the trade-off measurable, which is the only way to tell whether you priced the certainty correctly.
If you are quoting guaranteed rates off a blended average, that is worth rebuilding at adspin.io.
