Why Not All Paid Traffic Is Equal

In this week’s episode of Voices of Search, we spoke with Chester Scott, chief strategy officer at Lunio, about a problem most performance marketers know exists but rarely quantify: not all traffic hitting a paid campaign is worth the same. 

Chester’s background ranges from agency and platform to client-side marketing roles across the UK, Europe, the Middle East, and APAC, including an early stint at TikTok building out its e-commerce advertising vertical. Our conversation covered why click fraud is only the visible tip of a much larger waste problem, why traditional fraud detection is starting to break down in an agentic world, and why marketers who treat every signal as equally trustworthy are quietly sabotaging their own measurement systems.

Key Takeaways From This Episode:

• Roughly 8.6 to 8.7% of paid traffic is invalid or fraudulent, but total marketing waste—including low-intent users, non-incremental spend, and poor creative execution—climbs closer to 50%
• Fraud isn’t concentrated in the channels marketers assume; traditional search carries lower volumes of invalid traffic but isn’t “squeaky clean”
• The rise of agentic shopping is collapsing the old human-versus-bot framework that fraud detection has relied on for years
• Marketers have defaulted to optimizing for events—clicks, form fills, signups—that bots can replicate just as easily as humans
• Solving the measurement problem has to come before any AI-driven bidding system can be trusted with signal quality

The Waste Problem Hiding Behind Click Fraud

Chester opened by putting a number on something most marketers only sense anecdotally. Lunio—originally founded as PPC Protect before becoming a multiplatform traffic validation company in 2022—classifies paid traffic through proprietary detection algorithms, ingesting clickstream data across search and social platforms. 

Based on a study of more than two and a half billion clicks a year, the company estimates that roughly 8.6 to 8.7% of traffic coming from paid channels is invalid or fraudulent.

Click Fraud Is Just the Entry Point

That figure alone would justify closer scrutiny of paid spend. But Chester was clear that click fraud is only the visible layer of a much bigger problem. “The reality is that the waste problem beyond just click fraud and invalid traffic is much bigger,” he said. “We’re probably getting up towards 50% when you add in all of the additional waste vectors.”

Those additional vectors include:

• Low-intent users who click but were never going to convert
• Non-incremental investment—spend on traffic that would have converted anyway
• Poor creative execution that fails to qualify the right audience
• Audience saturation, where repeated exposure stops driving new results

Chester framed the scale of the issue with a line from John Wanamaker, the 19th-century retail pioneer credited with inventing the price tag: half the money spent on advertising is wasted, and nobody knows which half. The fundamental challenge hasn’t moved in over a century. 

What’s changed, in his view, is where the fix has to come from—not one dramatic overhaul, but what he described as sophisticated advertisers “trying to find marginal improvements in absolutely everything that they do in order to drive big effects when compounded together.”

Fraud Doesn’t Live Where Marketers Expect It

That framing set up a misconception Chester wanted to correct directly: the assumption that fraud clusters in a handful of “risky” channels—social, affiliate, blackbox systems like Performance Max—while traditional search stays clean by comparison. 

He didn’t dispute that those channels carry higher levels of invalid traffic. But he pushed back on the idea that established channels are exempt. “You do tend to see lower volumes of invalid traffic” in traditional search, he said, “but it doesn’t mean to say that it’s completely squeaky clean.”

Why the Risk Stays Invisible

The reason this misconception persists, Chester explained, comes down to how visible the incentive for fraud is. Fraud within social, shopping, or PMAX has an obvious logic attached to it. Search and LLM-driven discovery are different: “the incentives for fraud may feel a little bit more abstract,” he said. “They’re a little bit harder to get your head around.” 

That gap between visible risk and actual risk is exactly where budgets quietly bleed.

A Data Vacuum at the Worst Possible Moment

This blind spot is compounding at a moment when the channel mix itself is getting harder to read.

Chester broke the underlying data problem into three connected layers:

• Channel diversification: The mix marketers once managed—search, a bit of shopping, maybe Snapchat—has expanded into traditional search, social search, shopping, PMAX, and new agentic entrants from ChatGPT and TikTok
• Data volume: Years of chasing more data have given way to a harder question—separating what’s business-critical from what’s simply nice to have
• Measurement: Still largely unsolved, and built on attribution models Chester considers fundamentally flawed

None of these layers sits in isolation—more channels mean more data, and more data only helps if the measurement underneath it can be trusted. That third layer is where Chester spent the most time.

Why Last-Click Attribution Isn’t Enough

On that third layer, Chester was most direct. “Measurement, for the largest part, is still relatively unsolved,” he said. “The industry has maybe been a little bit complacent in accepting a fundamentally flawed approach in last-click attribution.” 

His preferred fix leans on causal proof, something closer to market mix modeling, paired with systems that can still guide day-to-day decisions. The tension is that causal proof tends to be a long-term indicator, while marketers activating campaigns daily need something that informs decisions in real time. 

“That’s the real challenge now,” he said, “how do we find a system that uses causal proof that’s able to balance the long and the short term?”

Why Agentic Commerce Breaks the Old Fraud Playbook

That measurement gap becomes more urgent given where consumer behavior is heading next. Chester was candid about the industry’s tendency to talk about shifts before they materialize. “We’re in a bit of an echo chamber,” he admitted. But he was equally clear the shift is underway. 

Agent-led traffic remains a small share of what’s identifiable today, “but do I think it’s going to remain a small amount? Absolutely not. That’s definitely going to grow exponentially.”

The Announcements That Moved Up the Timeline

Two recent product launches convinced Chester the shift is closer than most teams realize:

• Amazon’s Alexa for Shopping, which folds its on-site Rufus agent into a broader assistant experience
• Google’s Ask Advisor, unveiled at Google Marketing Live and Google I/O

Amazon’s move stood out to him in particular, given the company’s two-decade head start on consumer trust. “If there is a business that is able to get the consumer comfortable with leveraging agents on their behalf,” he said, “I think they’re probably the closest to it.”

From Human-or-Bot to Good-Bot-or-Bad-Bot

Fraud detection has historically relied on a simple binary: confirm whether a visitor is a real human, and you know whether the traffic has value. Agentic commerce collapses that logic. 

“We now need to understand what’s a real human, what is a bad bot, a fraudulent bot or a fraudulent user, and what is a good bot—a bot that’s maybe there with real human intent,” Chester said. 

His early answer is to shift the unit of analysis away from the source of traffic and toward intent itself.

Trust Will Be Built in Stages

Adoption, he cautioned, won’t happen as a single switch. He expects it to arrive gradually:

• Low-stakes research tasks first
• Finding flight or hotel deals next
• Making reservations
• Small transactions, then larger purchases

Each stage builds the confidence needed for the next. As Chester put it, consumers “have to have confidence that the agent is working in alignment with them” before they’ll hand over anything that matters.

The Trouble With Optimizing for Events

That same trust problem shows up closer to home, in how teams measure success. Chester pointed to a habit that’s become the default: treating raw events—clicks, form fills, signups—as the primary signal of whether a campaign is working. “It largely depends on optimizing for the right metric,” he said. “A click, a form download, a form submission—well, they’re all things that a bot itself can achieve.”

His fix isn’t to abandon events, but to tie them to an actual business outcome and validate the traffic feeding into that outcome first. He used paid versus organic search as an example: a team bidding on every available auction, regardless of whether that traffic is incremental, is paying for placements it may not have needed. “Do you really need to be in that auction? Probably not,” he said.

As bidding systems consolidate around AI-driven, black-box approaches, Chester expects the competitive edge to shift toward signal quality. 

“It’s all coming down to the data quality that we are ultimately able to build,” he said. Feed a bidding algorithm dirty data, in his words, and “those agents are going to be incredibly good at finding more and more rubbish.”

Validity, Not Volume, Is the Real Metric

Every problem raised in this conversation—click fraud, channel assumptions, agentic commerce, event tracking—traces back to the same root cause. Marketers keep optimizing for more: more clicks, more channels, more data, more automation. Chester’s argument is that scale was never the bottleneck. Validity was.

Teams that win from here won’t be the ones spending the most or moving fastest into the next channel. They’ll be the ones who stopped assuming every click, every signal, and every event is telling them the truth—and built the discipline to check.

Voices of Search is a daily SEO and content marketing podcast hosted by Jordan Koene and Tyson Stockton. The show delivers actionable strategies and data-driven insights to help marketers navigate the ever-evolving world of search engine optimization and content marketing. New episodes air weekly, covering everything from technical SEO to AI discovery, featuring industry leaders and practitioners sharing real-world frameworks and proven tactics.

Subscribe to Voices of Search on Apple Podcasts, Spotify, or your favorite podcast platform. Follow Previsible on LinkedIn for updates and subscribe to the VOS YouTube channel for video episodes and clips.