Cody Greco, CTO and cofounder at Prescient AI, loves working with great humans and building technology that saves them time and money.
Most people grapple with multiple competing responsibilities, commitments and distractions throughout any given day. There’s a reason life hack articles always suggest completing simple tasks immediately—because you never know when you’ll circle back to address them. For marketers, it’s critical to figure out how to fit into these fractured lives of potential customers. While your product may sound exactly what they need, there’s a significant chance your ad will catch them at the wrong time. A purchase may be put off by carpool duty, a work deadline or another pressing situation, but to say your ad failed would be short-sighted.
Unlike direct effects, through which the impact of ads is immediately visible based on clicks and sales, advertising “halo effects” occur when marketing activities indirectly impact something else downstream. Also known as advertising spillover, halo effects can amplify (or, in rare cases, diminish) the effectiveness of marketing campaigns in ways that are not immediately apparent but are incredibly useful to understand. For example, if someone sees an ad on TV and then searches for the brand on Google, the TV ad should still get credit for the resulting revenue.
Despite their angelic name and potential to squeeze additional value from media investments, halo effects have long been a devil for marketers. Everyone knows that consumers encounter ads in various places, but the inability to accurately account for halo effects has plagued ad measurement systems and hindered media optimization for decades. Because measuring their influence is so tricky, halo effects often go ignored.
When digital marketing exponentially increased the complexity of the advertising industry, multi-touch attribution (MTA) was introduced to address halo effects by assigning appropriate “credit” to each touchpoint leading to a purchase. Twenty years later, MTA has mostly resulted in disappointment. Piecing together comprehensive buyer click paths was never a simple task, and MTA fails to account for unlimited external variables that impact how consumers behave.
Here we are, in 2024, and many marketers are again (or still) relying on last-touch reporting that the ad platforms themselves have admitted needs to be revised. I’ve yet to meet any marketer who says the performance data they get from platform reporting matches their sales data. In most cases, the revenue amounts each channel claims credit for will add up to more than the total revenue for the period. Even using last-click data to optimize campaigns on a single channel is risky. It’s common for prospecting campaign halo effects to drive revenue that gets wrongly credited to bottom-of-the-funnel efforts, causing brands to oversaturate retargeting and miss out on customer acquisition potential.
Sometimes, backing up is the best way forward, but reverting to the comforts of last touch doesn’t go back far enough. Measuring halo effects has always been possible, not by following people around the internet, but through modeling. Analytics experts started using statistical models to quantify the impact of marketing activities as far back as the ’50s. Many proven techniques can measure the impact of halo effects. (If you really want to go down that rabbit hole, I discuss a few of them at length here.)
For decades, marketing mix modeling (MMM) has proven valuable for identifying patterns in performance and providing high-level insights into marketing effectiveness over longer periods. But, it wasn’t built to deliver campaign-level insights for daily or even monthly optimization. Due to the intricate ways marketing channels and tactics influence each other, halo effects are extremely complex to model accurately. To do it at scale, across all data inputs, at a speed that would be useful to marketers, has been impossible—until now.
Seemingly overnight, artificial intelligence has become omnipresent in our daily lives. It’s built into the latest operating system on every device and “powered by AI” is the term du jour to market anything connected to the internet. Enthusiasts and critics predict that generative AI (GenAI) will significantly disrupt how we do almost anything. However, this current fanfare barely scratches the surface of the possibility of this technology, which is still in its infancy. There are many subsets of artificial intelligence, and in my opinion, GenAI is the most limited of them.
I’m biased because I love math, modeling and statistics, but I’d bet on machine learning (ML) as a sleeper in the AI game. Recent advancements in ML are transforming the MMM we know from 70 years ago into an influential tool marketers will rely on to predict the future. Advanced ML models can deliver deeper insights faster than ever before, and the ability to finally understand halo effects provides a new view into the complex interplay of marketing activities and their impact on consumer behavior.
The Hurdles To Clear
These new models are based on extremely complex, advanced levels of math that will be highly protected by the teams and companies that build them. Marketers who are excited by the opportunities may find it challenging to vet MMM providers properly. Without knowing who’s doing ML modeling well, there are bound to be some letdowns. If technology providers don’t find some way to prove the effectiveness and accuracy of their innovation without spilling proprietary information, they risk never living up to their potential for lack of customers. That’s a genuine hurdle; the pressure is on to clear it before the excitement wanes. Those of us working in ML feel that pressure acutely because we see the opportunity most clearly—advertising that respects privacy but doesn’t kneecap marketers.
Once we find a way to earn the industry’s trust that the technology is sound, this new generation of MMM has the potential to drive a seismic shift in how brands plan, buy, measure and optimize advertising.
Using AI-powered MMM won’t write your copy or generate ad creative. But, it can inform you where to put your money to generate more revenue. For brands investing heavily in advertising, that’s a big deal.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
