First-Price vs Second-Price Auctions: How Ad Pricing Changed in Programmatic

In a first-price auction, you pay exactly what you bid. In a second-price auction, you pay $0.01 above the second-highest bid. The entire programmatic advertising industry switched from second-price to first-price between 2017 and 2019. This article covers how both models work, why the switch happened, what it did to CPMs on both sides, and how to adjust your strategy.

How Did Second-Price Auctions Work?

In a second-price (Vickrey) auction, the highest bidder wins but pays just $0.01 above the second-highest bid. You could bid your true valuation without risk. The auction mechanics protected you from overpaying.

Worked example: five DSPs bid on an impression. Bids come in at $4.00, $6.50, $7.20, $9.00, and $12.00. The $12.00 bidder wins but pays $9.01. The gap between their bid ($12.00) and their payment ($9.01) is $2.99 in savings they didn't have to earn through strategy. The auction gave it to them automatically.

This model dominated programmatic from its origins through 2017. The theoretical advantage was honest bidding: since you never paid your actual bid, you had no incentive to bid below your true valuation. In academic auction theory (William Vickrey won the 1996 Nobel for this), second-price auctions produce efficient outcomes because buyers reveal true willingness to pay.

The problem was that the theory assumed a single clean auction. Programmatic didn't work that way.

Why Did the Industry Switch to First-Price?

Header bidding broke the second-price model. Before header bidding, publishers called demand sources sequentially in a waterfall. One auction, one set of bids, second-price rules worked as designed.

Header bidding changed that by running multiple SSP auctions simultaneously for the same impression. Each SSP ran its own second-price auction internally, then submitted its winning bid to the publisher's ad server. The ad server compared these already-reduced second-price bids against each other, effectively running a second price reduction on prices that had already been reduced once.

The fix was straightforward: make every auction first-price. One bid, one price, no hidden reductions.

Index Exchange moved to first-price in March 2017. AppNexus (now Xandr) followed in 2018. Google AdX, the largest ad exchange globally, completed its transition to unified first-price auctions in September 2019. Once Google moved, the shift was effectively universal. By early 2020, second-price auctions in programmatic were functionally extinct.

How Do First-Price Auctions Change the Math?

Under first-price rules, the $12 bidder pays $12. Same impression. Same competitors. $2.99 more than they would have paid under second-price. That delta goes directly to the publisher (minus exchange fees).

The immediate impact was a CPM increase for publishers. Studies from the transition period showed publisher eCPMs rising 10 to 30% in the months after exchanges switched to first-price. Index Exchange reported a 20% eCPM lift for publishers in the first year.

For advertisers, effective CPMs rose correspondingly. If you didn't change your bidding strategy, you paid more for the same inventory. The same bid that cleared at $9.01 under second-price now cleared at $12.00. A 33% increase in cost for the identical impression.

This is where bid shading enters. DSPs developed algorithms to estimate fair clearing prices and automatically reduce bids to avoid overpayment. Bid shading brought advertiser CPMs back down, typically recovering 15 to 30% of the first-price premium. But it never fully closed the gap. Net advertiser CPMs after bid shading remain 5 to 15% higher than they were under second-price for equivalent inventory.

How Do Floor Prices Behave Differently Under First-Price?

Under second-price rules, a floor price acted as a phantom second bid. If the floor was $5 and only one bidder submitted at $14, the winner paid $5.01 (the floor + $0.01). The floor protected the publisher's minimum while second-price mechanics kept the actual payment low.

Under first-price, floors work differently. If the floor is $5 and a bidder submits $14, the winner pays $14. The floor prevented bids below $5 but did nothing to moderate the payment above it. The floor's protective function for publishers stayed the same. Its moderating function for buyers disappeared.

This changes how publishers should set floors. Under second-price, aggressive floors (set close to expected clearing prices) worked because they raised the effective second price without scaring off high bidders. Under first-price, aggressive floors risk blocking bids entirely since DSPs running bid shading may submit reduced bids that fall below a floor set too high.

The optimal floor strategy under first-price is dynamic: adjust floors based on historical clearing data by hour, geo, device, and ad format. Google Ad Manager supports rules-based floor configuration. Third-party tools like Assertive Yield and PubStack offer algorithmic floor management. Static floors leave value on the table in both directions.

What Should You Do Differently Under First-Price Rules?

If you're an advertiser: your DSP's bid shading is doing most of the work. Check whether it's enabled (it should be by default on The Trade Desk, DV360, and Xandr). Review your bid shading savings reports to understand how much reduction the algorithm applies by publisher and format. If you see consistent win rates above 30%, your shading is likely too aggressive and you're winning only low-competition auctions. Win rates of 15 to 25% suggest healthy calibration.

If you're a publisher: the switch to first-price was a net positive for your eCPM. Protect that gain by implementing dynamic floors and monitoring clearing prices regularly. If your clearing prices cluster tightly around your floor, the floor is doing the work. If they spread 40 to 60% above your floor, you're leaving money on the table by not raising floors or using dynamic optimization.

Check your current performance against CPM benchmarks by format and geo. If your eCPMs sit below benchmark despite decent traffic quality, floor configuration is the first place to look.

Use the CPM calculator to model the revenue impact of floor adjustments before implementing them live.