What is RGSP? Google’s Randomized Generalized Second-Price ad auctions explained

The concept of randomized generalized second-price (RGSP) auctions sent shockwaves through the PPC community after the subject took center stage at the Google antitrust trial.

While some digital marketers agreed with Google that the practice provides a better user experience, others sided with the Department of Justice (DOJ), arguing that it makes ad auctions unfair and purely helps line Google’s extremely deep pockets.

But what is RGSP, why does the DOJ think it’s problematic for ad auctions and how exactly does it impact Google’s ad revenue? Here’s everything you need to know.

How does Google pick an ad auction winner?

Dr. Adam Juda, Google’s Vice President of Product Management in Search Ads Quality Systems, explained at the federal antitrust trial:

The highest bidder doesn’t automatically win the ad auction.

A campaign’s long-term value (LTV) is instead given more weight.

This means Google sometimes loses out financially in the short term.

When advertisers bid on keywords, instead of determining an ad auction winner purely by bid amount, Google uses a metric called Ad Rank to decide how and if your campaign should rank. This collective score is calculated by examining:

Bid amount

Auction-time ad quality (including expected click-through rate, ad relevance and landing page experience)

Ad Rank threshold

Competitiveness of an auction

Context of a search query

Expected impact of assets and other ad formats

Your Ad Rank is recalculated every time time your campaign becomes eligible to compete in an auction, meaning your ad’s ranking may vary each time depending on competition, quality and search context.

Campaigns that don’t meet Google’s minimum Ad Rank threshold are automatically eliminated from the auction.

How does Ad Rank work?

Imagine five advertisers competing against each other in an ad auction with the respective Ad Rank scores of 80, 50, 30, 10 and 5. For this particular auction, Google requires a minimum Ad Rank threshold of 40 to rank above organic search results. This means that only the first two campaigns (with scores of 80 and 50) are eligible to show above organic search results.

In this instance, for an ad to be shown below organic search results, Google requires a minimum Ad Rank of 8. This mean that the campaigns with Ad Rank scores of 30 and 10 would qualify.

However, the campaign with the Ad Rank score of 5 does not meet the minimum criteria to appear above or beneath the search results and so will be eliminated from the auction, as shown in the table below:

Why doesn’t the highest bidder win?

If the highest bidder automatically won every Google ad auction, there is a risk the search engine could be left serving poor-quality ads. Poor-quality ads may not be relevant to a searcher’s query, which would likely result in a poor:

Click-through rate.

Conversion rate.

User experience.

This would decrease the overall value of Google’s product.

A Google spokesperson explained in a statement:

“Overall, higher quality ads typically lead to lower costs and more advertising success.”

“The Google Ads system works best for everybody when the ads we show are relevant and closely match what customers are searching for.”

It is also in Google’s best interest to serve high-quality ads that satisfy user intent because advertisers only pay Google when someone clicks on your ad, visits your site or calls your business.

Issues with Ad Rank

Without RGSP, ad auctions can get “messy,” as Frederick Vallaeys, CEO of Optmyzr, explained in How out-of-order ad promotion works on Google Search here on Search Engine Land.

An ad that meets all minimum criteria required by a Google auction can sometimes still rank below an ad that fails some criteria, he wrote. Vallaeys went further by using an example of an ad auction that had a 4% threshold for predicted CTR. The details of the competing bids are listed in the table below:

In the example listed above, Ad 2 meets the threshold because its predicted CTR is 5% – 1% higher than the 4% required by Google. However, because Ad 1 has a higher Ad Rank score (30), Ad 2 would rank further down the page to maintain auction fidelity, and only be displayed when Ad 1 is allowed.

“This is not a great scenario for advertisers or Google, so they address this by allowing ads to be shown in a different order than what ad rank would normally dictate,” Vallaeys wrote.

This “different order” refers to the concept of RGSP.

What is RGSP and how does it affect auctions?

RGSP is a practice leveraged by Google that picks the winner of an ad auction at random from the top bidders as long as their long-term values (LTVs – a Google calculation that is essentially the same as Ad rank) are close enough.

The top bidder then “pays the price of the bid equal to the next-highest bid plus one cent,” according to Big Tech on Trial.

The Department of Justice argued at the federal antitrust trial that this practice creates an unfair competition for bidding advertisers as the winner of an auction should always be the highest bidder.

Why is RGSP unfair?

Advertisers have two options if they want to avoid their potential winning bid from being demoted at random to runner-up:

Improve their campaign’s LTV.

Increase their bid amount.

The issue here is that Google hasn’t specified exactly how advertisers can increase their campaign’s LTV, which leaves them with one option if they wish to avoid RGSP – increase their bid amount.

To avoid RGSP, the bid amount would have to be significantly higher than the runner-up (as mentioned before, winners and runners-up can only be swapped via the RGSP process if the LTV and bid amounts are close enough). This has resulted in advertisers having to raise their bid 3.7 times higher, reports This Week In Google Antitrust.

What are the issues with RGSP?

Jay Friedman, CEO of advertising agency Goodway Group, highlighted the reasons RGSP could prove problematic for advertisers:

“Imagine you want to buy a ticket to a concert. Not everyone who wants a ticket can get a ticket, so there is an auction. You submit your bid and it’s not a first-price auction (highest bidder wins, pays what they bid,) and not even a second-price auction (highest bidder wins, pays a nominal amount [i.e. $1] over the second-highest biddger.) Instead, the concert venue holds an RGSP – a ‘randomized general second-price auction.’”

“Let’s say the the top two bidders submit bids of $100 and $95. In RGSP, the concert venue takes the top two bidders and, ‘as long as the long-term value of each of the bidders to the concert venue is pretty close,’ there is a chance the concert venue randomly swaps the top two bidders and awards the seat to the second highest bidder instead. Sounds like a deal if you randomly get the ticket for $95, and I guess frustrating for the highest bidder.”

“EXCEPT – the concert venue tells you there are two ways to make sure you don’t get randomly swapped out as the highest bidder. One, increase your long-term value to the venue. They don’t tell you how to do this and note it may include your behavior, referrals, your bid amounts, bid frequency, ‘and other bidder quality elements.’ You decide that’s pretty vague. The second is to increase your bid! And, as it turns out, you’d have to increase your $100 bid to $370 to get sufficient confidence you wouldn’t be outbid.”

What has the DOJ said about RGSP?

The DOJ has argued at the federal antitrust trial that rather than resulting in higher-quality ads, RGSP is being used by Google to boost ad revenue. In putting forward its case, the department shared an email Juda sent to his team at Google acknowledging the difficulty the search engine would have in selling this practice to advertisers. It read:

“[I]f I have to say, ‘[W]e randomly disable you if you don’t bid high enough,’ then I’m going to have another bad year at [Google Marketing Next] ;).”

There was debate at the trial as to what was implied by the use of a winking emoji in Juda’s message.

Does RGSP increase Google’s revenue?

Google vice president and general manager of ads Jerry Dischler testified at the federal antitrust trual that while he was unsure if RGSP resulted in advertisers increasing their ad auction bids, he could confirm that the practice increases Google’s ad revenue.

Dischler went on to tell the court that the search engine “frequently” changes the auctions it uses to sell search ads, increasing the cost of ads and reserve pricing by as much as 5% for the average advertiser. For some queries, the tech giant may have even raised prices by as much as 10%. However, Google tends “not to tell advertisers about pricing changes.”

The Department of Justice shared an email sent by Dischler back in 2018 to highlight the pressure his team were under to meet revenue targets given to Wall Street by Ruth Porat, Google’s Chief Financial Officer. In the documents, he claimed his team were “shaking the cushions” to increase revenue. He wrote:

“If we don’t meet quota for the second quarter in a row and we miss the street’s expectations again, which is not what Ruth signalled to the street, so we will get punished pretty bad in the market.”

“I care more about revenue than the average person but think we can all agree that for our teams trying to live in high cost areas another $100,000 in stock price loss will not be great for morale, not to mention the huge impact on our sales team.”

Is RGSP new?

Practices like RGSP are not new. In fact, Yahoo! gave an interview to The Register back in 2010 explaining it had been using “squashing” and second price auctions since 2007 to increase revenue. Yahoo!’s then chief economist, Preston McAfee (who now works for Google as a Distinguished Scientist) told the publication at the time:

“When someone has a really high ad click probability, they’re very hard to beat, so it’s not a really competitive auction. So that they don’t just win [every auction], we do squashing. This makes the auction more competitive.”

“The bidders respond by bidding higher. The one who was destined to lose is now back in the race, so they bid higher trying to displace the number one, and the number one is trying to fend them off so they bid higher too.”

“We can make the competition a bit more fierce using squashing, even on keywords where there’s not much bidding.”

McAfee did not confirm how much squashing Yahoo! does but did say it was constantly changing and “resetting the parameters”.

What has Google said about RGSP?

Google uses RGSP to prevent a bias where one winner takes all, Juda said at the federal antitrust trial. Commenting on the practice, he told the court “we flip [the winners of auctions with runners-up], otherwise Amazon always shows up on top,” Bloomberg reported.

Another reason for selecting winners at random is so that advertisers don’t need to worry that they may be bidding too much in ad auctions, which could result in them constantly feeling the need to adjust their bid amounts, according to Google.

With RGSP, the price advertisers pay is determined by the bid amount put forward by the next highest-ranked bid. Juda described this method as “advertiser-friendly.”

How has the PPC community reacted?

The concept of RGSP appears to have divided the paid search community, with many criticizing the lack of transparency around it.

PPC specialist Vincent Norris wrote on LinkedIn:

“So much for giving advertisers transparency, right? What does this mean for advertisers? Does ad rank even matter? I personally hope that Google will get more than a fine and a ‘slap on the wrist’ for this.”

Mike Ryan, Head of Ecommerce Insights at Smarter Ecommerce, commented

“Is this behavior ethical? No. Whatever the initial motivation, this is auction manipulation that appears to harm Google’s competitor set and yield undue revenues by increasing costs for everyone else.”

Tyler Jordan, digital marketing expert, added:

“All digital marketers need to be aware that Google’s bid auction is no longer an auction. A real auction’s outcomes are dictated by supply and demand, but we just learned from the horse’s mouth there are other factors in play.

“I’m sure Google is working on making its advertising product more effective – at least for retail advertisers, since that was certainly the focus of Dischler’s quotes. What I’m sure of is that continuing to game its own system at the expense of its customers is not a long-term strategy for success. I’ve got plenty of ideas for ways Google can get better for #b2b advertisers if they’re open to a more honest way forward.”

Google Ads expert Kirk Williams posted:

“Google, we genuinely love the product you first introduced. We are the ones who had told clients for years why Search is one of the best marketing channels of all time (what incredible marketing intent there is in a search term!!). Stop the money grab and start rebuilding trust. For the sake of the industry. Please! 🙁 #ppc #googleads #adwords #ecommerce #RGSP.”

However, others have argued that “out-of-order” ranking changes can help improve the user experience, as pointed out by Vallaeys:

“While out-of-order promotion changes the typical auction dynamics, Google believes it ultimately improves the search experience, and I tend to agree with that. For advertisers, it highlights the need to focus both on bidding strategically and optimizing for relevance and Quality Score.”

Content marketer Goutham Veerabathini shared this notion, commenting on LinkedIn:

“The introduction of randomness might help create a more dynamic and unpredictable auction environment to prevent strategic strict deterministic ranking of bids that gives top position always to only one player who bids the highest after mastering all the other factors.”

Why we care. Fair ad bidding is essential for advertisers to achieve their marketing objectives efficiently, maintain trust in the advertising ecosystem, and foster long-term relationships with advertising platforms and publishers. It contributes to a healthy and competitive marketplace where advertisers can optimize their strategies for better outcomes.

Deep dive. Read our Google antitrust trial updates for all the latest developments.

The post What is RGSP? Google’s Randomized Generalized Second-Price ad auctions explained appeared first on Search Engine Land.