Are Artificial Price Floors The Next Iteration Of Ad Fraud?

jim-carusoData-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.

Today’s column is written by Jim Caruso, senior vice president of product and client strategy at Varick Media Management.

The digital advertising industry has a history of concerns around viewability, brand safety and nonhuman traffic, especially within programmatic ad buyingWhile these are the most common factors that can compromise brand integrity, buyers face other types of fraud that often go unnoticed.

The rise of automation is closely linked with a perceived lack of oversight, leading to less scrupulous sellers gaming the system. Transparency issues related to pricing – specifically the price flooring within programmatic auctions – can go on for weeks before machines detect them. Could deceiving the buyer in this way be considered fraud as well?

While price flooring isn’t regulated, most buyers participating in auctions expect a second-price auction model. With this model, the highest bidder wins but pays the second-highest price rather than the amount they bid, which is usually a penny more than what the other bidder has offered. For example, a buyer who bids $8 over a buyer who bids $3 will expect to pay $3.01, or a price close to that number. If the buyer sees CPMs closer to $8 or more, suspicions arise.

Buyers understand that price floors are necessary to ensure that inventory brings in money for sellers and publishers. Buyers will often place very high bids on desirable inventory, assuming that they will pay far less for an impression when they win the bid. However, when the value of an impression is falsely represented, the trust between buyer and seller is quickly lost.

It’s worth noting that price flooring is OK on some inventory, when the impression is worth it and there is a high likelihood to convert. But it’s not OK when a supply source artificially sets a floor price on every single impression, including those that are available through other supply sources without the same floor. When this happens, buyers have a right to question whether this is a pure accident or if it’s a manipulation designed to raise the inventory supplier’s revenue.

Recently, traders spotted instances where a $30 bidding strategy on quality inventory won the auction, but the price ultimately paid was the $30 CPM high bid, rather than a price close to the second-highest bid.

To avoid a similar type of situation, advertisers must arm themselves with the necessary knowledge and technology. They can start by familiarizing themselves with artificial price flooring and requesting that their programmatic partners conduct periodic artificial flooring analyses to detect possible incidents.

Advertisers should utilize a team of dedicated and experienced traders to quickly identify this practice. Whether in-house or a third party, a human trader will immediately notice something is wrong when CPMs close at alarmingly high rates, whereas a machine could take far longer to detect the anomaly.

Advertisers can also request access to log-level data, a string of dozens of data points on every impression served across the Internet. Once an advertiser has initial suspicions, they can take a deep dive into their programmatic partners’ log files to see bid and close prices. This allows the advertiser and technology partners to decide which sellers to continue buying from by weeding out those with artificial price floors.

Finally, advertisers can maintain direct, personal access to the seller. The current anti-fraud practice is to simply turn off low-quality sellers. But in the case of artificial price flooring, the best approach is to directly reach out to the seller. Since artificial price flooring isn’t necessarily linked to low-quality inventory, advertisers may find themselves dealing with sellers who hold a lot of status in the market. They can discuss the price floors, find out if they are specific to the advertiser and inquire about the possibility of removing the floors.

When it comes to programmatic, an advertiser’s top priority is to deliver the best results while protecting its brand, whereas the publisher’s top priority is to optimize its yield. But to what extent are publishers doing this?

Programmatic is about efficiently paying fair prices for inventory, but in order to do so, advertisers must be aware of artificial price floors. The increased use of technology to transact media has made advertisers susceptible to errors and manipulation. Fortunately, it’s also made it easy to act swiftly to successfully combat these issues.

Follow Varick Media Management (@VarickMedia) and AdExchanger (@adexchanger) on Twitter.

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!


  1. Jim Metzler

    Perhaps the solution is in the third paragraph. Most buyers expect a second price auction model but if you check the exchange contracts, the pricing model is usually not spelled out.

    Artificial soft floors cause the auction to take on first price characteristics — the more you bid the more you pay without necessarily changing your win rate. Very aggressive soft floors turn the auction into a defacto first price auction.

    Perhaps the solution is to bid as if it really is a first price auction. You will need to track your bids and win rates to determine your optimal bid amount but it is doable. More complicated and harder to get right, but doable. Once you start bidding as a first price bidder, you control what you pay and the exchange cannot make you pay more.

  2. Justin Gospodarek

    This is certainly just a matter of perspective, which you started to touch on. Programmatic is about efficiency, sure. It’s built on top of a technology which has one purpose: to match supply and demand. This would be a massive stretch of the word fraud, I think. I’m from the agency world – I’d love a magic tool that always gets me low rates. That said, if I’m waving money around (in your example, bidding $8), I fully expect someone to try and take all of it. As the buyer, I know it’s my responsibility to figure out what the inventory is worth to my clients, and develop my strategy accordingly. I see no foul on the seller side for trying to maximize revenue when the demand exists.

  3. Great article Jim, I have been looking for this discussion in a while and could not find traction nor interest. I do believe this is already happening at scale and we are lacking the knowledge to understand and fight it.

    While the Vickrey based auctions seems the ideal “academic solution” for a real world problem, and we would probably agree during the creation of the RTB protocol, I suspect we never would have imagined this problems.

    I have seem different variables contributing to increase the complexity and issues, I would like to add :

    1 – Some SSPs allows for the publisher to have a Dynamic Price Floor, thus preventing the DSP to learn the static floor ( by sending different responses to each bid ) and capturing more money for the Publisher ( which seems fair to the Publisher ) . In practical terms, only the SSP is able to calculate the Delta and could easily skew that.

    2 – In Brazil (where I live) the currency for Media is Brazilian Reais and the exchange rates, changes every day. RTB bids are US Dollar only. First there is no single currency exchange rates for all players. Second, every day my price floor would be different ( example : $2.5, $2,44, $ 2,55 etc… ) This allow for more room in “artificially” frauding pennies/fractions per transactions and it’s a design flaw as we cannot require from the ends( buyer/seller), that the intermediaries all agree to the same rules for exchange rates.

    3 – I have heard of a real true story that a Buyer wrongly set CPC targets as the Campaign Budget ( blame AdOps and real time systems) meaning instead of a sub-dollar CPC it got a several thousand dollars CPC target. That caused the DSP to spend a few hundred thousand in seconds. Strangely the same targets used for this campaigns, never, never had such a high clearing BID price before they have bidded like that. Meaning that the behavior of the 2nd price increased together with the AdOps error ? too much of a coincidence for me.

    Until we have a way to compare detailed, individual BID x Clearing data with all parties involved in a neutral interest party, the middle-man will hold all control, responsibility and risks .

    Specially in OPEN type auctions, where the number of companies ( or players ) are so big that this seem not practical. But for PMP, checking back with your clients ( from a Publisher point of view ) could help to shed some light if this is already happening with you.

    Let’s keep the conversation,

  4. One of the things that would definitely help advertisers and buyers get to the bottom of this is if bid requests included whether or not it was being sold based on first or second price and whether or not an impression has a “dynamic” floors or ghost bids enabled.

    That can help inform better bidding strategies to either target or avoid artificially inflated inventory.

  5. Ad Fraud really? Thats a lack of over site by the advertiser if they are bidding at a rate they are not willing to pay for the inventory that is on them. Programmatic doesnt mean you dont have to do any work or actually pay attention to what you are bidding. The advertisers have the responsibility to control what is spent and where. It is far from Ad Fraud its due to lack of over site and attention to detail if anything.

  6. I’m going to have to agree with Brandon. Advertiser’s not being savy to the programmatic market doesn’t qualify as fraud. Publisher’s you could argue have received a raw deal in regards to the automation taking place in the industry. It has greatly lowered the value of their inventory while advertisers reap the benefits. The supply side wouldn’t stay in business very long if they weren’t putting effort into increasing revenue. By the time their inventory is bought and each technology partner has gotten their hands on it, they’re only receiving a fraction of it’s initial value.

  7. I think people are bringing valid arguments, to the wrong point.

    Fraud as the current definition by TAG ( as a reference, but not in a exhaustive way) is related to fake audience/results being paid. We tend to agree to that and fight that .

    What Jim Caruso is bringing in this article, can be really different. Let’s examine this section :

    “It’s worth noting that price flooring is OK on some inventory, when the impression is worth it and there is a high likelihood to convert. But it’s not OK when a supply source artificially sets a floor price on every single impression, including those that are available through other supply sources without the same floor. When this happens, buyers have a right to question whether this is a pure accident or if it’s a manipulation designed to raise the inventory supplier’s revenue.”

    So if a Publisher has a impression available trough several sources, it’s expected that different sources have different algorithm and strategies to win the monetization race.

    For me, Fraud could happen if some of these partners is increasing the floor without giving full transparency/sharing the profits back to the owner of the impression. And that can easily happen when you’r daisy chaining several SSPs, mixed with AdNetworks demand seeking the wrong goal* of selling every impressions. And this is quite common to see.

    Some could disagree with me and say this is plain classic arbitrage, real time arbitrage.

    It’s not going to take long for us to learn with Nasdaq and high frequency trading, the space is already there, and I bet it’s already working.

    * wrong goal in my personal oppinion, as demand is limited and supply is nearly infinite in the web, seems a wrong goal.