Causal Marketing Turns … Casual

Leaders at Casual Precision discuss why they’ve shifted gears to become a causal marketing mix model attribution business — and what they can do for brands and agencies with their new tools.

Brett Charest founded Casual Precision as a performance and direct-to-consumer media agency in 2015. In the 11 years since, he’s recruited a powerful team of leading experts: managing partner James Vassallo arrived in 2017; executive vice president of client growth Lissie Perkins came aboard in 2021; and chief technology/product officer Kevin O’Reilly joined in 2025.

In recent years, though, Charest and his team began to focus their performance media expertise on a bigger target than just creating successful campaigns for clients: how could Casual Precision create a more precise attribution model for an industry struggling with cross-channel conflicts and latent over-attribution?

Using Bayesian marketing mix modeling (MMM) and improving AI and machine-learning technology, Casual Precision has created a service that can help brands and agencies understand the impact of marketing spend before investing. And it’s a tool that speaks to each person in a campaign’s value chain, from the agency media buyer to the brand CFO and everyone in between.

Recently, Results caught up with Casual Precision’s leaders to talk about its shift, what it can deliver to clients, and where it sees performance media — and its own business — heading in the coming years.

Note: responses have been edited for space, style, and clarity.

Casual Precision has shifted its focus from agency to attribution, leaning into causal MMM using Bayesian modeling. Why this transition and why now?

Brett Charest: We pivoted because two things happened at once.

First, the core signal layers that direct response was built on are collapsing from the traditional mechanics like toll-free numbers and promo codes to IP address regulation, iOS updates, and walled garden platforms kind of grading their own homework. Second, traditional agency reporting has really started to become glorified accounting, telling you what happened after the money was spent rather than why it happened and where the next dollar should go.

Because our leadership team spent years running campaigns in the trenches, we built a platform that we desperately needed: a single environment unifying the micro tactical tools with a macro-Bayesian marketing mix model (MMM). Ten years ago, building a proper Bayesian MMM was a six-figure, six-month project. Today, we can stand one up in weeks, refresh it continuously, and put it in front of a client in a live platform.

Brands and agencies don't need another passive dashboard. They need an active decision-making engine. “The vendor's dashboard said it worked,” is no longer an answer that a CFO accepts in an environment where every media dollar must be defended.

What does, “Move beyond attribution, measure causality,” mean to you — and what should it mean to your clients and prospects?

Charest: Standard attribution is basically a bookkeeping exercise. It takes conversions that already happened and divides credit among whichever ads are nearby, thereby letting the bottom-funnel channels or baseline organic demand take full credit.

Branded search is a classic example. Basically, attribution gives Google search credit for a customer who was already standing in line to buy. Causality asks the real CFO questions like, “What would have happened otherwise? If we turn off this specific channel or cut it by 30 percent, what do we actually lose?”

For brands and agencies, the practical difference is simple. Attribution tells you where credit landed. Causal measurement tells you where the next dollar should go. Those are frequently opposite answers.

The channels with the best last-touch numbers often deliver the least incremental value, while broad reach channels like video and audio carry the whole growth system.

We're not asking clients to throw out their attribution stack. We're asking them to stop using it as a budgeting tool.

What is Casual Precision doing to help brands and agencies migrate into the complex ecosystems of modern performance media by using media science capabilities?

James Vassallo: Performance media hasn't just shifted. Everything's become much more layered. Linear TV, radio, direct mail — they haven't gone away. And we've got search, display, and email that have added that click-to-follow for decades now.

Now we have CTV, we have OTT, we have digital audio, and social that are really just adding more screens to that same person's movement across in one site session. It's more than just a shift. It's the layering of more and more screens to make decisions with.

Really, what we've done is taken our media experience and our expertise, and we've built this attribution platform from that reference. It's not just another channel with its siloed measurement tool that's creating its own homework — and usually over-attributing.

Brands look and see, “Facebook says this, TikTok says that.” By their logic, they're technically right, but the numbers don't add up. What we've really created is a platform that's the referee to weigh every channel against a brand's actual KPIs, and it comes with credibility and predictability.

It starts with our Bayesian MMMs that evaluate total business outcomes, macro trends, and channel interactions to help set the optimal budgets for clients — whether it's for the next month or the next quarter. What helps that model is our event-level multi-touch attribution — Prism — that's designed to understand site engagement and how you got there. Then we have standardized tools included within the platform, like spot attribution and direct attribution, which are included for in-flight tactical decisions like creative mix, network mix, network optimization, and placement optimization.

Within our platform, what's most important is that we are calibrating our modeling efforts on a continuous basis by testing. We are putting together calendars to help brands and agencies lay out local incrementality testing for channels like linear TV or radio, and we're helping them to build specific audiences for which we can design a digital media test and control group so that we continue to keep educating our model.

The end result is that agency-side leaders — media buyers, analysts, strategists — and brand-side positions — acquisition growth managers, CMOs, CFOs — are all reviewing the brand KPIs from a central point. With all those positions come different profiles from which to view the platform, so each can get the answers to the questions that they're asking. A CFO is asking a different question than the brand-side or agency-side media team. So, we've created specific profiles to cater to those positions.

The Casual Precision team, led by Brett Charest and James Vassallo (front and center with K1 object), celebrated 10 years in business last year and continues to thrive together.

Charest: Today, we see CFOs that are looking marketing spend as an investment. So, we want to show them as an investment.

This is an investment tool. Your market dollars are an investment into your business. It's not an expenditure. In order for them to view it as an investment, they need to have the confidence in the modeling that when they enter a question — “I want to cut TV by 30 percent. What's that going to do?” — they’ll get a reliable answer. “Here's what your short-term gain is going to be, and here's what your long-term gain is going to be.”

Vassallo: A simple example that comes up across multiple profiles or multiple positions is, “Did TV work?” A media buyer can go in and look at the TV spot attribution dashboard and find which stations and which day parts responded the best and decide where next week's spots really should get optimized. The analyst can look an incrementality test that we're running and look at the geo test readout and say, “Hey, I am confident of the lift that we saw within this five-market test against this control basket.” Then, to Brett's point, you’ve got your CMO or your CFO saying, “TV's marginal return at the current spend looks good. What happens if we shift TV the budget up 20 percent?” We can give a solid prediction of what the outcome would be.

How does the “Halo Effect” from broad-reach channels still drive success in performance campaigns?

Lissie Perkins: Broad reach channels like TV and radio drive performance success by building lasting brand equity that lowers acquisition costs across all channels over time, particularly downstream digital channels. Our marketing mix model will look at all your marketing, and it'll attribute short-term sales. It has a short memory.

It's looking at ad stock decay over weeks, maybe months, but it's completely losing the long-term momentum that top-funnel media generates. That's why we built Halo, which is an extension to our MMM. While the base model will handle that immediate response, Halo will quantify how top-funnel media fuels performance down funnel over time.

Crucially, it extracts that long-term payback from the total return. So, it's not stacking on top. It's not double counting. This gives marketers an honest, true long-run impact of their channels so they don't accidentally starve the engines driving growth.

How is Casual Precision helping break brands’ addiction to “last click” attribution by isolating incrementality and proving cross-channel correlation?

Perkins: Last click is addictive because the platforms are right there at your fingertips. You log in, you see your sales, you hit your CPM, you say, “Good!”

But when an ad network takes credit for any click or any IP match, they're ignoring true incrementality. Just because a user saw an ad and then converted doesn't mean that that ad actually caused that conversion. Often, it would have happened regardless.

So true incrementality requires that we have a holdout group and that we measure the exposed and the unexposed groups for isolated lift — and we design these real-world experiments for the media type. So, for addressable media like CTV or streaming audio, we run holdout groups or PSA control groups, and we measure that user lift on the exposed vs. unexposed.

But when we're talking about broad-reach media or even black box media — TV, radio, and YouTube — we run geographic testing where we have exposed markets measured against matched control markets so that we can prove local lift. So, when the CFO then sees and says, “My holdout group is converting at the same rate as my exposed group,” that last-click myth evaporates.

What is Casual Precision doing to address the siloing of attribution and resulting signal loss for its clients?

Kevin O’Reilly: From an optimization perspective, if you're continually investing into a channel that is showing performance at “X” because it's taking credit from other channels, then you're going to keep on putting more and more money into it thinking you're getting returns.

And because returns aren't instant, you may not see that loss of opportunity or actual profitability until maybe two or three weeks down the path, or even from a Halo perspective, five or six months down the path. To be brutally honest, at the heart of over-attribution in silos, it really comes simple myopia, right? Facebook doesn't know that you're on television. TikTok doesn't know that your competitors just spent an awful lot of money and drove category search. And your email doesn't look like it's getting the right capture because somebody, instead of clicking on a link, went outside and did either a brand search or a category search.

Customer journeys don't start and end in a channel. With the level of fragmentation that we're starting to see, you've got to be capturing these population behaviors so you can actually model them. The modeling that we built is basically population modeling. It's Bayesian. At some level, it’s Gaussian distributions across media spend, media impressions, media clicks, and as you bring these values in and then factor in ad stocks— branded search is a really short ad stock; TV has a longer ad stock — you're going to make sure that the credit's getting a better opportunity to drift back to the channel that created it.

And then you start to look at how it drives bottom of funnel engagement. We can take a look at different parts of the customer journey, and then we can take it back and compare it to what the silo said. In some instances, you're seeing as much as 85-percent over-attribution from a channel. A lot of the time, that's because they're doing one of two things: obviously, they're robbing other media, or — most of the time — they're actually robbing the baseline.

I care about your media. I care massively about your internal marketing activities. And I care about your primary externalities. All of those things are going to drive behavior. The reality is we're spending into demand cycles.

So how do we tease those out? How do you make sure the channels are getting the appropriate credit that they deserve going through top, bottom, and through the mid-funnel to ensure that you something in front of you that is a simulation for your business to understand what the outcomes are likely to be before you commit to them.

Casual Precision built its new MMM system with data privacy in mind. With personally identifiable information (PII) essentially removed from the data mix, agencies and brands can execute campaigns without worrying about running afoul of data safety issues.

What does Casual Precision do to help its clients navigate — and mitigate — compliance risks, turning client data into a low-risk, high-reward opportunity?

O’Reilly: The good news about aggregate models is that they’re personally identifiable information (PII) free. The data that we bring in is daily trend level data: spend, impressions, orders, visits, etc. It's aggregate pretty much by design — that's PII free.

Now, when you go to our pixel solution, we've actually bent over backward to design it to be privacy-first in the way that we collect data. So, you put our pixel in your app or on your website, and what comes through the system and ends up being stored is identity free. It has an anonymized key next to it.

That anonymized key is the same key if you have us plan the campaign digitally. So, it's on both sides. What you've kept is your direct attribution, such as CTV attribution, etc., but you don't have any privacy markers. We lean into put the pixel on your website, so you're not sending over user logs or naked IP addresses. Closing all that down stops the potential for human error. Any engineer worth his salt can install it in 10 seconds because we do all the heavy lifting. It's simply a pass-through.

How is Casual Precision using hybrid tracking technology to provide neutral measurement of true incremental ROI for its clients?

O’Reilly: We believe that the most important thing and the greatest value that we have to provide to our clients is a comprehensive understanding of how their media activates their target consumer base. We want to make this type of modeling and this type of behavior analytics available to pretty much anybody through the ecosystem. That's the starting point: we don't necessarily manage your data.

But we're going to take our expertise from managing data and embed it in the platform. For example, “This is how you build a testing control. We'll do it for you.” It'll activate in the platform itself. It'll tell you exactly what markets you're supposed to go after.

Spot attribution doesn't just do the usual in this system. Users will see a presentation screen of what's working and what isn't. We have two different new algorithms in it. The first one advises on how you change your media plan — here's what you buy, and here's what you minimize. And the new model goes one step further. It breaks down your attribution into all of its component parts and says, “Here's what you should be buying that you're not. Here's the combination of day parts and content and genre and creatives that you should use to then test.” We lean very heavily into, “You're here, but you want to get to this next place.”

Every test you do feeds right back into the marketing mix model. They're informing the model on a continual basis about what is changing. We also measure externalities — CPI, affordability — to try and help understand how that's shaping things. That's what we mean by neutral. The platform just delivers hard truth. It doesn't sugarcoat it.

The other place that we lean into neutrality is on your data management. One of the biggest problems that you have in any marketing mix model is touching data. It's dirty. It's often late. It creates an awful lot of person-to-person communication. And all of those things reduce the amount of time that you're going to get your insight.

Traditionally in marketing mix models, insight comes way too late. What you want to make sure you do is to build data management to get the information into the platform as fast as possible. So, where we can use authorization tools, we use them. If you use our algorithm, we’re using AI-based processes we built that automates that whole process. So instead of it being a hands-on solution, it's mostly hands-off and data now flows continuously into the attribution models.

Minimizing data management gives the marketer more opportunity to respond to the events that need to change. If you go back 10 years and tell people you could optimize your television media, by network, by station, people would have looked at you like you have 16 heads.

That's now standard. The next new standards for optimization are always on marketing mix models, always on test-and-learn models, always on spot attribution and CTV models.

It’s one platform, but it's many different modeling methodologies. I'm not trying to answer everything with a single model. It's many methodologies inside the platform that give you the ability to answer the question that you need to answer as the CFO, as the media planner, even as a data scientist.

What do you believe is Casual Precision’s key differentiator in its competitive set?

Charest: Three things set us apart.

First, it’s measurement by media experts. We aren't just software engineers selling a black box SaaS product and walking away. We are media practitioners who built an end-to-end operational engine that handles the unglamorous heavy lifting, data normalization, post-log ingestion, and pixel management.

Second, we offer a full scope suite in one environment. We handle the entire time horizon in one platform from the immediate micro tools like spot attribution, direct attribution, and MTA to midterm Bayesian marketing mix models and to the long-term brand equity modeling across all channels.

And then three, accessible enterprise economics. Traditional enterprise modeling required giant consulting teams and static annual readouts. We've automated the analytical pipeline so clients can get a continuous living model with scenario planning and an AI-powered portal at a fraction of the cost.

What do you think Casual Precision’s position of strength is today? And where do you see it being by 2029?

Vassallo: Today, our strength is giving enterprise brands, direct-to-consumer brands, and performance agencies this platform. It’s got credibility, and we win in a room where a CMO and a media team are asking different questions because our models of transparency are built people with decades of experience and through proper testing.

It's built by people that who really understand media and marketing execution. We pride ourselves on having the experience in media and marketing execution to build the platform.

Charest: By 2029, all measurement will stop being static postmortem reports and become active business simulations. Marketers will sit in our platform, run live flight simulations, and ask questions like, “What happens to net revenue if I move $500,000 from linear TV to CTV next month?” They’ll get an instant answer back with statistical certainty. Planning, buying, and causal optimization will collapse into a single continuous loop.

And Casual Precision is built to be the default operating system powering that unified future for the entire media ecosystem.