What It Takes to Know What an Auction Cost
The durable answer to the title's question is that CPM is a unit or summary, while auction cost is an outcome of coupled allocation, pricing, delivery, and measurement decisions. A defensible cost claim must therefore identify the population being measured, trace the bid through the auction's rules, and state where the evidence ends. That standard applies whether the auction uses a familiar second-price label, a closed-loop ranking model, or a proposed mechanism such as the Information-Bundling Position Auction.
Cost is a path, not a single field
A CPM can be perfectly accurate and still be an incomplete explanation. It may describe the amount an advertiser entered, the effective value used to rank that offer, the clearing price charged to a winner, or an average calculated after delivery and billing records were reconciled. Those statements refer to different stages. A report that uses the same label for them has not necessarily made an arithmetic error; it has created an interpretation problem.
The path starts with an impression opportunity and ends with a measured outcome. A search, site visit, app session, or streaming event can create an opportunity to serve an ad. The seller makes that opportunity available under inventory and business rules. Advertisers submit bids through buying software. The auctioneer decides which demand is eligible, how qualified bids are valued and ranked, whether a reserve is met, who receives the placement, and how the winner pays. A delivery or measurement system then records what was served and aggregates the financial result.
The Media Rating Council (MRC) calls allocation and pricing the two core functions of an ad auction. They must be considered together. Allocation tells us which opportunity was assigned and to whom; pricing tells us the transfer required for that assignment. If allocation changes, the population behind a reported average can change before the payment rule does. If pricing changes, the same allocation pattern can produce a different financial result. If measurement changes, the reported CPM can move even when the auction decision is stable.
That is why four questions should remain separate:
- What was offered? The nominal bid is the advertiser's entered amount, expressed in a unit such as CPM, CPC, CPV, or CPA.
- What competed? The effective bid is the value the auctioneer attributed to the nominal bid after applicable factors and the qualified demand set were considered.
- What was charged? The clearing price is produced by the auction's payment rule, which may be first-price, second-price, modified second-price, or another disclosed formula.
- What was reported? The delivered or average CPM summarizes a defined set of measured impressions, costs, adjustments, and time windows.
The distinction is useful even when the numbers happen to match. If a buyer enters a $10 CPM and the final reported average is also $10, equality does not prove that the nominal bid was used unchanged, that every auction cleared at $10, or that the same denominator would produce the same average tomorrow. It only shows that the visible quantities are equal under the definitions used in that report.
The opposite mistake is equally common: treating every gap between the numbers as evidence of a hidden charge. A lower reported average may reflect a post-auction discount, a different delivery population, or fewer high-pressure opportunities. A higher average may reflect a reserve, a first-price payment, stronger competition, a changed supply mix, or a fee treatment. The gap tells us that stages differ; it does not identify the cause without the relevant records.
A defensible cost claim has six conditions
The MRC's Digital Advertising Auction Transparency Standards provide a practical way to turn this distinction into evidence. The framework does not prescribe one auction format, require publication of precise proprietary algorithm details, or claim that every platform currently exposes every field. It asks auctioneers to disclose enough about methods, variables, outcomes, changes, and errors for participants to understand what they paid and how the price was determined.
1. The unit and population must be explicit
Before comparing CPMs, identify whether each number is a bid, a clearing price, an invoice amount, or an average over delivery. Then identify the denominator: auction opportunities, won impressions, served impressions, viewable impressions, or another measured population. Add the time window, supply scope, placement, and gross-or-net billing treatment.
This is not clerical detail. Suppose a campaign is paced and wins a selective set of opportunities. Its average over delivered impressions may be lower than the price it would have faced if it entered every opportunity. Another campaign may enter a narrower but more competitive supply set and report a higher average. Comparing the two as if they were identical markets turns a denominator difference into a false pricing conclusion.
The same care applies to allocation. A reserve can leave some opportunities unsold, and a qualification rule can remove bids before competition. A report based only on sold impressions says something about the transactions that cleared. It does not by itself describe the value or cost of every opportunity that reached the seller.