4 questions you should ask to catch programmatic ad fraud

Fraud risks are growing more sophisticated. The ANA’s Q2 2025 Programmatic Transparency Benchmark found that $26.8 billion in global programmatic media value is lost each year to supply-chain inefficiencies, fraud, and low-quality inventory. 

Fraud is especially prevalent in connected TV (CTV) advertising. One of the fastest-growing channels in digital, CTV has become the newest playground for bad actors who know detection has not kept up with spend.

Most marketers mistakenly assume that the biggest threat is sophisticated fraud itself. It’s not. The biggest threat is the passive assumption that someone else — a demand-side platform (DSP), an agency, a publisher partner — has fraud handled. 

That assumption is how budgets leak. Not in a single blown buy, but steadily, campaign after campaign. Surfacing fraud doesn’t require a deep technical background, just the ability to ask the right questions and press for real answers. 

Here are four questions every programmatic advertiser should be asking. The answers get harder, and more important, as campaigns move from display to video to CTV.

Your customers search everywhere. Make sure your brand shows up.

The SEO toolkit you know, plus the AI visibility data you need.

Start Free Trial

Get started with

1. Where is my ad actually running?

Before you can spot fraud, you have to understand where your ad ends up. Programmatic ads travel through a chain of middlemen, exchanges, resellers, and verification layers before they reach a publisher. That chain is called the supply path, and the more hops it has, the harder it is to know what you actually bought.

On display, your ads run on websites and inside apps. Ask your partner to walk you through the supply path for your campaigns. Is your inventory coming directly from trusted publishers, or is it moving through a long chain of open-exchange resellers? 

Display is also where page-level quality signals matter most. The industry is making real progress here, with new tools that surface granular page-level signals like ad density, refresh rate, and content quality. That kind of transparency is the direction every display advertiser should push their partners toward.

On video, ask the same supply-path question and add another: What kind of player loaded the ad? 

An autoplay in a muted corner of a page isn’t the same impression as a pre-roll on a premium video property. Both can end up on your invoice at similar prices. The same page-level signals apply to video running in web environments.

CTV is a different animal, as CTV ads don’t run on pages. They run inside apps: publisher apps like NBC and ESPN, streaming platform apps like Hulu, Max, and Paramount+, and free ad-supported streaming apps like Pluto TV and Tubi. 

The question shifts: Which app claimed the impression, and was it really that app? Fraudsters spoof both devices and apps, running impressions on server farms that pretend to be Roku or Fire TV devices within a premium streaming app. 

Ask your partner specifically what app-level and device-level verification they’re running, and which apps make up the bulk of your spend.

2. Who is verifying the traffic?

Bot detection on display is mature. Independent verification vendors have spent years tuning their filters, and baseline traffic hygiene on most DSPs is reasonable out of the box.

Video is trickier because video ads live in far more varied environments. Vendors can reliably track viewability, but catching sophisticated bots that mimic real viewing behavior takes more work.

CTV is the soft spot. I’ve watched this pattern before across two decades in programmatic: Every time a channel scales faster than detection can keep up, fraudsters follow the money. 

You can probably guess what the CEO of a CTV company will say about this one, but the numbers are the numbers. DoubleVerify’s 2025 Global Insights Report found that bot-driven activity accounted for 65% of CTV fraud, with an estimated four million infected devices generating fake traffic daily.

Fewer independent vendors have reliable CTV detection, and the bad actors know it. Don’t accept a general reassurance that “fraud is filtered.” Ask specifically what your partner’s CTV fraud detection looks like, which vendors are doing the filtering, and what share of your CTV impressions got removed last quarter. 

If no one can produce that number, you’re not being measured, you’re being told a story.

Recommended Articles

How to wreck your brand in 30 seconds

Feb 17, 2026

How to use CRM data to target the right B2B audiences

Feb 17, 2026

The case for and against CTV in 2026 B2B marketing budgets

Jan 5, 2026

3. Is someone actually looking at the placement data?

Most marketers don’t want to live inside placement reports, and that’s fair. Granular reporting is dense, and parsing it is often one of the reasons you work with a partner in the first place. 

Still, you should know what your partner is looking at, how often, and what they flag when something looks off.

On display, site-level placement data should be a baseline output. If your partner is handing you summaries without naming where your impressions ran, press for more detail or ask them to walk through it. On video, publisher and placement reporting can get lost in vague categories. Ask for it by line item.

Staying current matters. Earlier in 2025, low-quality sites built to capture ad dollars were the industry’s top placement concern, with hundreds of millions of dollars in open exchange spend still flowing to those domains. 

The industry responded. Better detection, tighter inclusion lists, and shared research from firms like Jounce Media drove that exposure down to a fraction of what it had been. 

But the threats shift. Jounce’s mid-2025 research found that unnecessary resale layers in the supply path have now overtaken low-quality inventory as the primary source of waste in programmatic. 

That finding ties directly back to the supply-path question in section one: The more middlemen there are between you and the publisher, the more budget leaks out before your ad reaches a real person.

On CTV, app, and network breakouts are frequently aggregated in ways that hide the real inventory composition. You can’t spot these issues in a summary. You need a partner who is pulling the details and flagging what doesn’t belong.

4. How curated are my direct deals?

A quick definition: The open marketplace is the broad, anyone-can-bid layer of programmatic where inventory trades in real time. Direct deals and private marketplaces are pre-arranged buys accessed through a deal ID. The assumption is that a deal ID means vetted, quality inventory. 

That assumption deserves scrutiny.

On average, private marketplaces run more cleanly than open exchanges. The ANA’s ongoing Programmatic Transparency Benchmark confirms that. 

But averages hide a wide range. The same research shows that the least disciplined advertisers still see significant exposure to low-quality, made-for-advertising inventory, even within their curated deals, while the most disciplined run nearly clean. 

The difference isn’t the deal ID, it’s what goes into the curation behind it.

On display and video, ask your partner what actually goes into their private marketplace deals. Are they using page-level quality signals and third-party data to screen inventory, or are they bundling whatever the publisher offered and calling it curated? 

With CTV, direct deals with major networks, streaming platforms, and free streaming apps are still the cleanest path, but the same rule applies: A deal ID with no curation behind it is a label, not a guarantee.

The ANA’s most recent benchmark, published in February 2026, found that top-performing advertisers converted nearly 57% of their programmatic spend into fraud-free, viewable, quality impressions. Lagging advertisers converted less than 38 percent. That gap isn’t a measurement quirk. It’s the price of real curation.

Why disciplined advertisers win

Fraud-proofing isn’t a one-and-done setup task, but an ongoing discipline. Start with an allowlist of vetted sites, apps, and networks built from curated deals from partners doing real, page-level quality work. Treat blocklists as a reactive backstop, not a strategy. Watch your own incentives, too: Cheap CPMs buy you the cheapest impressions. 

When a performance metric looks too good, treat the spike as a signal, not a win. The advertisers who consistently ask these questions are the ones who know what their media is actually worth. This strategy applies to display, video, and, above all, CTV.