How does MMM measure halo effects across sales channels?

A street musician playing outside a busy subway station has one obvious way to judge a good night: count the bills in the open guitar case. That number misses a lot, though. Someone who loved the set but didn't carry cash might look up the band on the train and stream the whole album. A commuter who caught two songs might buy tickets when the band plays a local venue next month. The guitar case shows what the performance earned on the spot, and it says nothing about everything else that happened because of it.

Marketing campaigns work the same way. Ad platforms count what lands in their own guitar case, which is the clicks and conversions they can connect back to one of their ads. The shopper who saw your Meta ad and bought on Amazon two days later doesn't land there. Neither does the person who searched your brand name after a YouTube spot or picked up your product at a store. When budget decisions run on the guitar-case number alone, brands end up cutting campaigns that are doing far more work than they get credit for, and that's an expensive mistake to repeat every planning cycle.

Key takeaways

• A halo effect is revenue a campaign drives indirectly through another channel or sales destination, like Amazon, branded search, direct website traffic, or retail stores.
• Ad platforms can only see the conversions they can connect to their own ads, so they routinely miss halo effects and sometimes count the same sale more than once.
• Marketing mix modeling estimates halo effects by comparing spend and impression patterns over time against several revenue streams at once, while accounting for delayed responses and decay.
• Not every MMM measures halo effects, since many traditional models treat channels as independent and predict a single revenue total, which pushes spillover into baseline demand.
• Prescient's marketing mix model measures halo effects at the campaign level across revenue destinations, including nonpaid Amazon revenue influenced by channels like Meta and Performance Max.
• Halo effects can come from any campaign, not only top-of-funnel ones, so cutting a campaign based on platform-reported ROAS can shrink revenue somewhere else.
• Halo effect estimates tell you how much revenue a campaign likely drove in other places, but they can't name the individual shoppers who moved between channels.

What a marketing halo effect actually is

A halo effect is the revenue a marketing campaign drives indirectly, through a channel or sales destination other than the one where the ad ran. If a campaign's impact spills over into Amazon sales, branded search, organic search, or direct traffic to your site, that spillover is its halo effect. We think the campaign deserves credit for that revenue, even though its ad platform will never report it.

Halo effects aren't only an upper-funnel story, either. Awareness and prospecting campaigns tend to produce more of them, but a retargeting ad that someone sees without clicking can still lead to an Amazon purchase later that week. Any campaign can create revenue somewhere it can't see.

Cross-channel effects vs. cross-platform sales effects

It helps to split halo effects into two types, because they show up in different places in your reporting. The difference comes down to whether the sale still happens on your own storefront.

Cross-channel effects usually lead to misplaced credit, since the revenue is visible but assigned to the wrong source. Cross-platform sales effects are harder to spot because the revenue often never gets connected to any marketing at all.

Direct response, halo effects, and baseline demand

Any sale your brand makes can usually be traced back to one of three sources. Separating them is the whole job of halo effect measurement, so it's worth being clear on the definitions.

The trouble is that halo effects and baseline demand look almost identical in a dashboard. Both show up as Amazon sales, organic traffic, or direct visits with no ad attached. Without a measurement approach that can tell them apart, revenue your campaigns created can easily get filed as revenue that would've happened anyway.

Common halo effects marketers overlook

Some halo effects are so common that most marketers have seen them in their numbers without having a name for them. These four come up again and again for omnichannel brands, and each one shows up in a place the original campaign's ad platform can't see.

Meta advertising influencing Amazon purchases

A shopper sees your product in their Instagram feed and heads to Amazon to buy it, maybe for Prime shipping, the reviews, or a checkout they already trust. Unless that shopper clicked a specially tracked link, the Amazon sale doesn't get connected back to the Meta ad. Meta reports a weaker campaign, and Amazon reports a nonpaid sale that seems to come from nowhere.

YouTube campaigns contributing to branded search

YouTube ads often reach people when they're busy watching something else, so a lot of viewers don't click. Some of them remember your brand name, though, and search for it days or weeks later. Your branded search campaign then captures that sale and looks like a star performer, while the YouTube campaign that created the demand gets little credit.

Influencer marketing increasing direct website visits

People who hear about your brand from a creator they trust often skip the link and type your URL straight into their browser. That visit shows up as direct traffic, which looks like brand loyalty or baseline demand. Promo codes catch some of these sales, but plenty of shoppers never use the code, so the creator campaign's full impact stays hidden.

Digital advertising influencing physical retail purchases

For brands sold at Target, Walmart, Ulta, or other retailers, a meaningful share of digital ad impact can end in a store aisle. Someone sees your ad on their phone and grabs your product during their next shopping trip because it's right there on the shelf. These retail halo effects are some of the hardest to see, since there's no click, no pixel, and no shared ID between the ad and the register.

Why platform reporting rarely captures the full halo effect

Ad platforms are pretty good at one thing: counting the conversions they can connect to their own ads. That's useful, but it's a narrow view by design. A few structural limits keep platform reporting from ever seeing the full picture.

Platform-specific attribution windows and identifiers

Every ad platform sets its own rules for what counts as a conversion, usually a click or view within a fixed window that it can match to a user through cookies, device IDs, or logged-in accounts. A shopper who converts outside that window, switches devices, or blocks tracking drops out of the count, even if the ad did its job.

Different conversion environments

Meta can't see inside Amazon's checkout, and Google can't see what happens at a Target register. Each sales destination keeps its own data, so a sale in one environment that was influenced by an ad in another falls through the gap.

Double-counting and missing credit

When a shopper clicks a Meta ad and later a Google ad before buying on your site, both platforms may claim the full sale. Meanwhile, the Amazon purchase Meta influenced gets claimed by no one, so your platform totals overstate some campaigns and understate others.

How MMM estimates halo effects

Marketing mix modeling takes a completely different approach from platform tracking. Instead of following individual shoppers, an MMM starts from what's observable, like the revenue that actually landed on your website, on Amazon, or at retail, and works backward to estimate what drove it. That's what makes halo effect measurement possible, since the model doesn't need a click or a pixel to connect a campaign to a sale.

Not every MMM measures halo effects, though. It takes specific design choices, which are worth walking through one at a time.

Marketing spend and exposure patterns over time

An MMM looks at how your spend and impressions for each channel or campaign rise and fall over time. When a YouTube campaign ramps up and branded search revenue reliably climbs a few days later, that pattern is evidence. Across months of data, with spend moving up and down at different times for different campaigns, those patterns let the model estimate how much each campaign contributes to each outcome.

Multiple sales outcome series

This is where a lot of MMMs fall short on halo effects. If a model only predicts one total revenue number, it can estimate how much a campaign contributed overall, but it can't tell you whether that revenue landed on your site, on Amazon, or through branded search. To measure halo effects by destination, the model has to track several revenue streams separately and estimate each campaign's contribution to each one.

Prescient's MMM does this. For example, it treats nonpaid Amazon revenue (Amazon sales that didn't come from Amazon Ads) as something outside marketing may have influenced, including Meta and Google's Performance Max campaigns. That lets a brand see how much of its Amazon revenue was likely driven by marketing running somewhere else entirely.

Lagged responses and carryover

People rarely buy the moment they see an ad, and ads keep working after they stop running, since their influence fades gradually instead of switching off. An MMM has to account for both the delay and the decay, or it'll miss revenue that arrives later and credit it to whatever happened to be running that day.

Many MMMs apply one fixed decay shape to every channel. Prescient's marketing mix model considers delayed effects and decay when estimating downstream revenue, and it lets those effects vary from campaign to campaign, because a CTV campaign and a retargeting campaign don't fade at the same rate.

Separating real relationships from coincidence

Two things moving together doesn't mean one caused the other. Your Meta spend and your Amazon sales might both climb in November simply because it's holiday season. A good MMM has to separate the revenue your marketing drove from seasonality, promotions, and baseline demand that would've shown up anyway.

This gets especially tricky because brands tend to spend more during the exact periods when demand is naturally high. When a model can't tell those apart, marketing's impact tends to leak into baseline demand, or the reverse. Prescient's mechanistic modeling approach is built to represent how channels interact with each other and how demand changes over time, which helps it keep those explanations separate.

Traditional MMM vs. Prescient's approach

Because halo effect measurement depends on these design choices, the type of MMM you use matters a lot. Here's how a typical traditional MMM compares with how Prescient's marketing mix model works.

The biggest difference is in that last row. When a model treats channels as independent and predicts one revenue total, spillover has nowhere to go except baseline demand, so it can look like your brand is just naturally strong when your campaigns are actually creating that demand.

How to use halo effect findings

Halo effect data is only useful if it changes what you do with your budget. The good news is that the most valuable moves are pretty straightforward once you can see total contribution side by side with what each platform reports.

Compare total contribution with platform-reported conversions

Start by lining up each campaign's platform-reported revenue against its modeled contribution across every destination. The gap between those two numbers is often where the most important budget insights live. Here's a hypothetical example of what that comparison can look like for a single Meta prospecting campaign with $50,000 in spend.

In this example, Meta slightly overstates the campaign's direct revenue while missing more than half of its total impact. A team looking only at the 1.2 ROAS might cut this campaign, even though it's one of the strongest in the account once halo effects are counted.

Avoid cutting campaigns that create demand elsewhere

The most expensive mistake halo effect measurement prevents is cutting a campaign that looks weak in-platform but feeds revenue everywhere else. Prospecting campaigns are the classic example, since their main job is creating demand that other campaigns or channels often capture later.

This also changes how you read retargeting and branded search, which tend to look fantastic in platform reporting because they capture demand at the last step. Prescient credits the campaign that created the demand, not just the one that captured it, which often shifts credit away from retargeting and toward the prospecting work underneath it. Prospecting campaigns still drive plenty of direct conversions too, so you'll want to see and quantify both.

Optimize toward total business outcomes

The goal isn't the highest ROAS inside any single platform. It's the most total revenue across your website, Amazon, retail, and every other place customers buy. Once you're judging campaigns on total contribution, budget questions get easier to answer, like where to scale, where to pull back, and which campaigns are propping up others.

A few quick checks can help you put that into practice:

• Flag any campaign where modeled contribution is much higher than platform-reported revenue before you cut it.
• Look at which destinations a campaign's halo effects land in, especially Amazon and retail, so your channel teams know where the revenue is going.
• Watch for campaigns that look great in-platform but have little modeled contribution, since they may be capturing demand other campaigns created.
• Plan budget shifts using forecasts that include halo effects, so a reallocation doesn't shrink revenue somewhere you weren't watching.

Prescient's Media Forecaster feature includes halo effects data when it forecasts what happens if you scale a campaign's spend or shift budget into it from other channels. That keeps brands from cutting prospecting campaigns that are driving indirect conversions.

What halo effect estimates can't tell you

Halo effect estimates are powerful, but they're estimates, and it's worth being honest about where they stop. Knowing their limits helps you use them well instead of expecting something they were never designed to do.

They don't identify the exact people who switched channels

An MMM works with aggregate patterns, not individual journeys. It can estimate that a Meta campaign likely drove $40,000 in Amazon sales last month, but it can't hand you a list of the shoppers who made those purchases. For budget decisions, that's usually fine, and it's a big part of why MMM holds up as privacy rules tighten, since it never needed to track individuals in the first place.

Correlated media and unobserved demand can introduce uncertainty

Some situations make halo effects harder to estimate. When several campaigns always scale up and down together, the model has a harder time separating their individual contributions. Demand shifts the model can't see, like a viral social post or a competitor stocking out, can also blur the picture.

That's why it matters to see how confident a model is in its results, not just the results themselves. Prescient shows a model confidence score alongside its findings, so teams can weigh how much to rely on a given estimate when making budget moves.

Where Prescient comes in

Prescient's marketing mix model was built to show what platform reporting can't: the full revenue each campaign drives across your website, Amazon, and the other places your customers buy. Our MMM measures halo effects at the campaign level, updates daily, and accounts for delayed effects and decay that vary from campaign to campaign. So you can see the gap between a campaign's apparent return inside its ad platform and its modeled contribution everywhere else.

That view helps omnichannel brands stop cutting campaigns that are creating demand and start funding them with confidence, using our Media Forecaster feature to plan budget shifts that account for halo effects. If you want to see how much revenue your campaigns are driving outside their own platforms, book a demo.