Your best-performing campaign might be cannibalizing your other channels

Every sales team has that one closer. They come in at the end of a long process—after months of calls, relationship building, and careful nurturing—shake hands, and log the win. Their numbers look exceptional. Their conversion rate is the envy of the team. But ask the account managers who did most of the work and the story gets more complicated. The closer didn't create that deal, they collected it.

Marketing channels work the same way. Some campaigns do the hard work of building awareness, generating demand, and moving people through a consideration journey they didn't know they were on. Others arrive at the moment of purchase and book the revenue. When every channel gets measured independently, the closers look like stars. The work that made them possible barely shows up in the numbers, and the budget follows the ROAS.

Getting this wrong doesn't just mean undercrediting some campaigns. It means misunderstanding what's actually driving growth, scaling the wrong things, and quietly eroding the pipeline that feeds the channels getting all the investment.

Key takeaways

• High ROAS in a single campaign or channel doesn't mean that campaign is generating demand. It may be harvesting demand that other campaigns created.
• Demand generation and demand harvesting are functionally different jobs with very different efficiency profiles, and standard attribution tools can't reliably tell them apart.
• A heavily scaled branded search or retargeting campaign can cannibalize organic traffic and direct visits, shifting revenue between channels without increasing it while appearing to perform exceptionally in isolation.
• Cross-channel interactions mean that the marginal impact of any one campaign depends on what else is running alongside it. Measuring campaigns as if they operate independently produces misleading results.
• Halo effects describe the positive version of this dynamic: awareness and upper-funnel campaigns driving lift in organic, branded search, direct channels, and sales in other marketplaces. Understanding halo effects and cannibalization requires the same kind of modeling.
• Marketing mix modeling (MMM) surfaces the statistical relationships between campaigns and revenue across the full mix, making it possible to distinguish genuine demand creation from demand shifting.
• Prescient captures halo effects at the campaign level, including revenue driven through branded search, organic traffic, direct visits, and retail channels, so brands can see the real contribution of every campaign in their mix.

Why high ROAS isn't the whole story

ROAS is a ratio: revenue attributed to a campaign divided by what was spent on it. Like any ratio, its meaning depends entirely on how the numerator gets calculated, and that calculation almost always has a hidden assumption baked in: that the revenue credited to a campaign was revenue the campaign produced.

That assumption often doesn't hold. A retargeting campaign that reaches people who were already close to converting will look highly efficient because it's meeting buyers at the moment of intent. But the intent was built somewhere upstream by a prospecting campaign, a CTV spot, an influencer post, or months of brand-building that never showed up as a direct click. Standard attribution, whether last-click or multi-touch, tends to assign credit to the most recent and most trackable touchpoint. The campaign that did the heavy lifting earlier in the journey rarely gets its fair share.

This creates a persistent pattern: the campaigns closest to conversion look best, get scaled, and attract more budget. The campaigns that built the conditions for conversion look marginal, get questioned, and attract less. Over time, the brand invests more and more in harvesting demand it's progressively less good at generating.

The demand generation vs. demand harvesting problem

Not every campaign is trying to do the same thing, and that's fine. A healthy marketing mix needs both. The problem isn't that demand harvesting campaigns exist, it’s that brands can  lose sight of which campaigns are doing which job (and often start optimizing purely for efficiency metrics that favor harvesters over generators).

A prospecting campaign on Meta or TikTok reaches people who haven't thought about the brand yet. Its job is to create a need, establish a memory, and start a consideration journey. It will rarely produce a high ROAS in the platform dashboard because the conversions it generates are diffuse, delayed, and often credited to whatever channel the person eventually converts through. A retargeting campaign, by contrast, reaches people already in the funnel. It has a much shorter path to conversion, and it shows well in any attribution model that follows click paths.

The efficiency gap between these two types of campaigns isn't necessarily a signal that the prospecting campaign isn't working. It's often a signal that it's working exactly as intended, just in a place the measurement isn't looking.

How channel interactions distort what you see

Things get more complicated once you account for how channels interact with each other, because they rarely operate as independent systems. When a brand's upper-funnel spend is healthy, it doesn't just drive impressions, it also raises branded search volume, increases direct traffic, and makes retargeting campaigns more effective because the audience they're reaching already has some familiarity with the brand. All of those effects show up in channels other than the one that created them.

The reverse is also true: when upper-funnel spend drops or disappears, the downstream channels don't immediately crater. They coast on the awareness that was built before. Branded search volume holds for a while, direct visits stay relatively stable, and retargeting keeps converting. And because the metrics look fine, there's no obvious signal that the pipeline is thinning until months later when everything starts to soften simultaneously.

This time lag between cause and effect is part of what makes cross-channel interactions so hard to see with standard reporting. By the time the measurement catches up, the connection to the original decision has been obscured by a dozen other changes in the mix.

The cannibalization scenario

The most pointed version of this problem goes beyond undercrediting upper-funnel spend. Some campaigns actively shift revenue between channels while appearing to produce it, and that difference matters for how you should think about scaling them.

A branded search campaign is a useful example. When someone who was already planning to visit the brand's website searches the brand name, clicks a branded paid search ad, and converts, the paid search campaign logs a conversion. But that visitor was likely coming anyway; the ad intercepted organic or direct traffic and redirected it through a paid channel. The brand didn't acquire a new customer, it paid to capture one it had already earned. At low spend, this might be worth it for control and visibility. Scaled aggressively, it can meaningfully inflate attributed ROAS on paid search while suppressing what would have been free organic revenue.

Retargeting creates a similar dynamic at scale. When a retargeting campaign is the dominant touchpoint for a large pool of people who were already actively considering the brand, it can look like an extremely efficient conversion driver. But many of those conversions would have happened through direct visits or organic search regardless. The retargeting campaign has moved revenue into its own attribution column, not generated net new revenue for the brand.

Neither of these things is inherently bad at reasonable investment levels. The problem is when the attribution model treats demand shifting as demand creation, because the budget decisions that follow are based on a fiction.

What it looks like to measure this correctly

The shift required here is from reading channels in isolation to modeling the relationships between them. An MMM tool observes how changes in one campaign's spend correspond to changes in outcomes across the entire mix, including the organic, branded search, and direct channels that don't have a spend line of their own. That's what makes it possible to ask whether scaling a retargeting campaign is actually growing total revenue, or whether it's redistributing revenue from channels that would have converted those customers anyway.

The same cross-channel modeling logic that surfaces halo effects—the positive spillover that upper-funnel campaigns create in downstream channels—also surfaces the scenarios where one campaign is absorbing credit from others. Both are versions of the same underlying reality: channels don't operate independently, and measuring them as if they do produces a distorted picture of what's actually working.

When brands can see these relationships clearly, the budget conversation changes. Instead of scaling whatever has the best ROAS number in isolation, the question becomes which campaigns are expanding the pie and which ones are just rearranging the slices. 

Where Prescient comes in

Prescient's MMM captures the full contribution of every campaign in a brand's mix, including the revenue that flows through branded search, organic traffic, direct visits, and retail channels as a result of campaigns that never got direct-click credit. That's what makes it possible to see both sides of the cross-channel story: the halo effects that show awareness spend driving downstream lift, and the scenarios where demand is being shifted rather than created. The model runs at the campaign level and updates daily, so brands aren't working from a point-in-time read that might be obscuring a pattern that's been building for months.

If your top-performing campaigns have never been examined in the context of what they might be taking credit for rather than creating, that's a conversation to have before peak season. Book a demo to see how Prescient maps the relationships across your full marketing mix.