Last Tuesday, while you were obsessing over your campaign’s rising CPMs, your biggest competitor was in their boardroom celebrating a 240% increase in engaged users. Their deck looked incredible. Charts trending up and to the right. Efficiency metrics that would make any CFO smile.
Here’s what they didn’t mention: their actual revenue dropped 18% during that same period.
They weren’t lying, exactly. Those users existed-sort of. They clicked. They engaged. They showed up in every dashboard and report. There was just one problem: they weren’t human.
Welcome to the $81 billion-per-year fraud tax on digital advertising, where bots have gotten so good at impersonating your ideal customer that they’re actually outperforming real people on the metrics you’ve been trained to care about.
The Metrics Are Lying to You (And You’re Paying Them to Do It)
I’m going to tell you something that might make you uncomfortable: there’s a decent chance that 20-30% of your “best performing” digital campaigns are feeding bot farms in Bangladesh, click farms in Moscow, and sophisticated fraud operations that look more like Silicon Valley startups than criminal enterprises.
But here’s the really uncomfortable part-your current setup is practically designed to reward this fraud.
Think about how most media buying relationships work. Your agency or partner gets compensated based on managed spend or hitting efficiency targets like cost-per-click or CPM. Now consider that fraudulent inventory consistently delivers impressions at $1.80 while legitimate inventory costs $3.50.
What behavior are you actually incentivizing?
A fintech company came to us after three years of “exceptional performance” with their previous agency. On paper, everything looked incredible. In reality, when we implemented actual fraud filtering and verification, their reported engaged user count dropped by 73%.
But their qualified leads increased 34%.
They’d been optimizing for metrics that bots could deliver more efficiently than real people. And they’re not alone.
Why Fraud Detection Tech Isn’t Saving You
The industry’s answer to this problem has been increasingly sophisticated detection technology. Machine learning algorithms. Behavioral analysis. Device fingerprinting. Pre-bid filtering.
All useful. None of it enough.
Here’s why: the fraud operators are running the same technology you are. Actually, they’re often running better versions because they’re testing against your fraud detection systems in real-time, at massive scale, iterating daily instead of quarterly.
Modern bot networks don’t behave like the clumsy spam bots of 2010. They’ve studied how real users move their mouse. How long people actually watch videos. What realistic browsing patterns look like across sessions. Some fraud operations are literally training their bots on the same datasets that legitimate companies use to understand human behavior.
The 2023 ANA study found that advanced bot networks now mimic human behavior so accurately that even trained analysts struggle to tell the difference. These aren’t script kiddies running operations out of their basement. These are well-funded businesses with R&D departments.
If your entire fraud prevention strategy is “we use verification vendor X,” you’re not preventing fraud. You’re just determining which fraud you’re sophisticated enough to detect this quarter.
The Supply Chain Is Fundamentally Broken
While everyone focuses on bot traffic, domain spoofing reveals just how compromised the programmatic ecosystem really is.
Your DSP tells you that impression was served on WSJ.com. Premium placement, premium price. Except it actually appeared on totally-not-fake-wsj-site.biz, routed through four intermediaries who each took their cut while maintaining plausible deniability.
The infrastructure that was supposed to prevent this-ads.txt files, supply chain transparency initiatives-exists. But compliance is voluntary, enforcement is weak, and fraud operators have gotten very good at working around the guardrails.
Remember when Uber’s former head of performance marketing cut $120 million in programmatic spend and saw zero impact on business outcomes? This opacity was a huge part of why. If you can’t verify where your ads actually run, you can’t verify much of anything.
The Attribution Scam Nobody Talks About
Bot impressions are one thing. Attribution fraud is where the real money gets stolen.
Here’s how it works: Your legitimate marketing efforts drive a high-intent user toward conversion. They’re going to buy. Maybe not this second, but soon. Sophisticated fraud detection systems can identify these high-intent signals. So just before that user converts, the fraudster serves an impression-often completely invisible-and claims last-click attribution credit.
You pay a premium for a conversion that was already happening. The fraudster gets paid for doing nothing. And because it shows up as a legitimate conversion tied to an ad exposure, everyone celebrates the performance.
I’ve audited accounts where brands were paying for the same conversion three or four times across different channels. Each claimed credit within their attribution window. Each was technically “correct” according to the platform’s reporting. At least one was complete fraud.
The simpler your attribution model-especially if you’re heavy on last-click-the more profitable it is for fraud operators to target you.
It’s About to Get Much Worse
Everything I’ve described is about to accelerate as third-party cookies die.
Right now, fraud prevention tools at least have consistent signals to track across the ecosystem. As we fragment into a landscape of alternative identifiers-Unified ID 2.0, Topics API, contextual targeting, first-party data strategies-every transition creates new gaps for fraud to exploit.
The brands building fraud-resistant infrastructure right now, before cookie deprecation fully hits, are going to have an enormous advantage. Everyone else will be scrambling to figure out what’s real while fraud operators exploit the chaos.
What Actually Prevents Fraud (Not Just Detects It)
After working with dozens of brands spending millions monthly on paid media, I can tell you exactly what separates companies that control fraud from those that subsidize it. It’s not about having the best verification vendor. It’s about completely rethinking how you structure, measure, and optimize digital marketing.
1. Redefine What Performance Actually Means
Stop optimizing for metrics that fraud can easily game. This means a fundamental shift in how you define success:
- Replace “cost per click” with cost per qualified lead or sale
- Replace “engagement rate” with downstream conversion contribution
- Replace “reach” with addressable audience quality scores
- Replace “video completion rate” with verified attention metrics
This doesn’t mean abandoning upper-funnel metrics. It means every metric you track should have a plausible connection to people who could eventually give you money. If bots can deliver it better than humans, it’s the wrong metric.
2. Take Control of Your Supply Chain
Supply path optimization isn’t just about reducing ad tech tax. It’s about eliminating the opacity that fraud requires to survive.
Work directly with publishers whenever possible. When you must use programmatic, demand complete transparency on every intermediary: every SSP, every exchange, every reseller. If a partner won’t provide full supply chain visibility, remove them. Period.
Yes, your reach numbers will drop. Your effective reach-the reach that matters-will increase. You’d rather reach 100,000 real people than 500,000 “people” where 300,000 are bots, right?
3. Layer Your Verification Like You Don’t Trust Anyone
Because you shouldn’t. Every verification vendor has blind spots because fraud operators specifically test against them.
Build multiple layers:
- Pre-bid blocking: Eliminate known bad actors before you even bid
- Ads.txt verification: Confirm supply chain authorization
- Third-party verification: Independent measurement of what was actually delivered
- First-party validation: Server-side verification of post-click behavior
- Business outcome correlation: Statistical analysis connecting campaign exposure to downstream value
If a campaign passes the first four layers but fails business outcome correlation, something’s wrong. Either your targeting is terrible or you’re buying fraud. Either way, kill it and reallocate.
4. Use Incrementality Testing as Your Fraud Detector
The most powerful fraud prevention tool isn’t a piece of software. It’s the scientific method.
Run holdout groups. Conduct geo-lift studies. Design proper incrementality tests. These methodologies answer the only question that actually matters: “Did this marketing cause business outcomes that wouldn’t have happened otherwise?”
Fraud, by definition, cannot survive incrementality testing. Bots don’t become incrementally more likely to purchase your product because they saw your ad. Real people do.
When clients push back on this approach because it’s “too complex” or “takes too long,” what I hear is: “We’re not ready to know the truth about what’s actually working.”
5. Fix Your Incentive Structure
This is the most important change and the one almost nobody implements.
If your agency or media partners get paid based on spend volume or managed spend percentage, you’ve created an incentive structure where expensive fraud is more profitable to them than cheap legitimate traffic.
Move toward outcome-based compensation tied to independently verified business results. Not platform-reported conversions. Not viewability rates. Actual business outcomes you can verify independently.
This is harder to structure. It requires more sophisticated measurement. Many agencies will resist.
Good. Their resistance tells you exactly how aligned they are with your actual goals.
The Competitive Advantage Hiding in Plain Sight
Here’s what makes this entire situation fascinating from a strategic perspective: every dollar your competitors waste on fraud is a dollar they’re not spending to compete with you for real customers.
Let that sink in for a moment.
If industry fraud rates sit around 20-30% (and that’s probably conservative), and you’ve implemented rigorous prevention getting you down to 5%, you effectively have 15-25% more budget than competitors spending the same amount.
But it compounds way beyond the math. While they’re training their algorithms on bot behavior, you’re training yours on real users. While they’re building lookalike audiences from datacenter IP addresses, you’re building them from actual customers. While their attribution models optimize toward fraud patterns, yours connect actual cause and effect.
Over time, this creates a strategic moat that’s almost completely invisible in standard competitive analysis. Your competitors can’t figure out why your campaigns consistently outperform theirs even though they’re spending more. They’ll blame creative, or audience strategy, or platform expertise.
They won’t realize they’re competing with one hand tied behind their back-the hand that’s busy high-fiving bots.
The Conversation You’re Avoiding
If you’re a CMO or marketing director, here’s the question that should terrify you:
“What percentage of our ‘successful’ digital marketing is actually just well-documented fraud?”
You probably don’t know the answer. Even worse, your organization’s incentive structure might be actively preventing you from finding out.
Those metrics you reported to the board last quarter? The performance that justified your budget increase? The campaigns that earned bonuses for your team? How much of it survives rigorous fraud filtering?
This isn’t about assigning blame. The programmatic advertising ecosystem was built with scale and efficiency as primary objectives, with trust as an assumption rather than something that needed verification. We’re all operating in a system that accidentally made fraud profitable and detection difficult.
But “we didn’t know” stopped being a viable defense somewhere around 2018.
Why Your Metrics Will Get Worse (Before They Get Better)
I need to be honest with you about what happens when you implement genuine fraud prevention.
Your reach will drop-sometimes dramatically. Your efficiency metrics will deteriorate. Your cost per impression will increase. Your engagement rates will fall. Every dashboard will look worse for a while.
Your executive team will ask uncomfortable questions. Your agency might push back. Your media partners will suddenly have very strong opinions about why fraud isn’t really that big a problem in your specific campaigns.
And your business will grow faster than it has in years.
Because you’ll finally be optimizing for something that matters: reaching real people who can become real customers who spend real money.
The brands willing to watch their vanity metrics temporarily suffer while they build fraud-resistant infrastructure are the ones that will dominate their categories over the next five years. Everyone else will keep celebrating dashboard green while wondering why growth is so hard to come by.
What We Do Differently
At Sagum, we built our entire approach around a principle that sounds obvious but is surprisingly rare: every dollar should drive real business outcomes.
That means fraud prevention isn’t something we bolt on at the end. It’s embedded in our strategy from day one. Multi-layer verification. Supply path optimization. Incrementality testing. Outcome-based success metrics. All standard, not optional add-ons.
We limit our client roster specifically so we can implement the kind of deep verification and testing that actually works. Because hitting your numbers with bot traffic isn’t hitting your numbers. It’s subsidizing a fraud industry while your real competitors capture your market.
When we take on a new client, we’re honest about what they’ll see: metrics that might look worse initially as we filter out the fraud their previous approach was buying. But within 60-90 days, business outcomes start improving. Not reported outcomes-actual revenue, actual customers, actual growth.
The question isn’t whether you can afford to invest in serious fraud prevention.
It’s whether you can afford to keep paying the fraud tax while pretending everything’s fine.
Your competitors are hoping you’ll keep celebrating those engagement rates from Bot Farm #4,782. We’re betting you’re ready for something real.
