Missed the live webinar, watch it here!
Written and contributed by: Thuy Ho
Fraud has always been one of programmatic advertising’s ugliest problems, but in 2025 the numbers are staggering: up to 22% of spend is lost to fraud each year. This results in lost trust between buyers, sellers, and clients.
That’s why, on September 3rd, Prebid hosted “The State of Fraud within Programmatic and Strategies to Combat It” webinar. The webinar was led by Christian Janelli and Prebid’s Buyer Task Force to unpack the scope of the problem and shared how open-source solutions can help the industry push back.
Prebid opened the webinar by restating its mission: to keep digital advertising free and open through interoperability across legacy and emerging channels. What began in 2017 with just 56 members has grown into a 200+ strong, fully member-led organisation by 2025. Prebid’s members shape the roadmap, write the code, and lead the conversation. Today, Prebid powers nearly 300 bid adapters, 40+ RTD modules, and 50+ analytics adapters, making it the clear leader in open-source programmatic technology.
Become a Prebid member to contribute to the market with us!
Why fraud is still draining budgets
Fraudsters have grown more sophisticated, slipping through basic verification and exploiting fragmented supply paths. By the time issues surface, the spend is already gone. The types of fraud are diverse: from bot traffic, click farms, and MFA sites, to domain spoofing, ad stacking, geo masking, and malvertising. Each tactic chips away at efficiency and erodes confidence in digital media.
How buyers can play a role
Buyers are spenders who also bring visibility into the whole marketplace. That perspective is critical for identifying risks and pushing the ecosystem forward. Prebid highlighted several ways buyers can contribute:
- Join working groups like the Buyer PMC, Video PMC, or Measurement PMC.
- Push for adoption of industry standards (ads.txt, sellers.json, schain).
- Demand log-level data access and transparency from partners.
- Prioritise supply path optimisation to cut out weak or fraudulent links.
Best practices against fraud
The webinar outlined a multi-layered defence strategy:
- Verification tools (HUMAN, IAS, DoubleVerify, Pixalate) to detect invalid traffic in real time.
- Pre-bid and post-bid filtering for viewability, bots, and anomalies.
- Data transparency with log-level analysis to spot suspicious patterns.
- Supply Path Optimisation (SPO) to reduce intermediaries and focus spend on trusted supply.
Prebid’s approach: moving from post-bid to pre-bid
Today, most measurement happens after the ad impression is served, which often leads to inefficiency and waste. Prebid is working to flip this model by passing standardised, verifiable signals pre-bid, enabling SSPs and DSPs to make real-time decisions before money is lost.
Benefits of this shift include:
- Cleaner supply paths and higher CPMs.
- Increased accountability with a potential Prebid “seal” of quality.
- Less reliance on middlemen, giving buyers and publishers more control.
- Reduced false positives on invalid traffic and brand safety, thanks to pre-bid tokens.
Q&A
- How are you objectively identifying brand safe, fraud-free, non-MFA traffic, and why should we trust the seal?
Goal is to collect signals pre-bid within Prebid auction time limits. For IVT, use MRC standards to identify and classify fraud, with clear taxonomy on why a decision was made. For MFA, surface signals like page structure and ad density rather than blunt block lists so buyers can decide. For brand safety, “objective” varies by brand; current systems are over-restrictive. Plan is to pass real-time context in the bid request so buyers make the call. Open source via Prebid and encrypted signals increase transparency and trust. - The cost per request seems expensive. Did you do a cost-benefit analysis?
Yes. Focus has been on collecting a minimal, useful set of pre-bid signals without impacting auction latency, staying within Prebid timeouts. Cost model aims for publisher pays once to transmit signals in all bid requests, keeping downstream costs near zero. Bringing it into open-source Prebid mitigates publisher skew because the signalling is transparent. - Is “AI slop” something to address with pre-bid signals?
Likely yes. It sits near IVT or MFA conceptually and should be identifiable via on-page characteristics. The current framework can support it if buyers prioritise it.
What about brand safety and suitability for in-app and CTV, especially with dynamic or updated content?
Not investigated yet, but issues are similar. In-app brand safety today is weak and often an honour system because most apps do not install multiple measurement SDKs. Prebid’s app implementations could enable a single communal SDK to collect signals once. CTV is harder without JavaScript; a Prebid-driven SDK approach is the likely pat