Adobe’s new Marketing AI let Customer cut ad spend 50% while growing Sales 10%

This post was originally published by Anonymous at Forbes (Innovation)

It’s a marketer’s dream. Maybe even a midnight fantasy: growing sales while cutting advertising and marketing spend. But Adobe’s marketing AI means that at least one massive cosmetics customer is actually living that dream.

And so could you, theoretically, with one little caveat.

Do you spend at least $50 million a year in advertising?

Adobe announced today that 80% of their marketing cloud customers are using AI to boost marketing ROI by personalizing customer experiences, creating better ads and emails, and accelerating processes. Director of product strategy Ali Bohra told me last week that Adobe is using AI to solve one of the toughest problems in marketing: knowing what drives sales. The framework is called media mix modeling (or market mix modeling) and it’s literally a 50-year-old technique that measures all campaigns, all paid media, all organic social and web activity — in short, all action — and tells marketers how effectively they’re doing their jobs in driving demand.

While old, media mix modeling is undergoing a revival of sorts because it is a macro-level methodology that is generally privacy-safe. In other words, it doesn’t necessarily rely on surveillance capitalism methods of tracking people and devices that government regulation and technological innovations are increasingly limiting.

But it does rely on scale.

“We’re measuring all campaigns, not just Adobe, not just paid media,” Bohra says. “We’re looking at the incremental value of each touch. Only Adobe has done that … only Adobe has enough scale in the market.”

Very likely marketing cloud competitors like Oracle, Salesforce, Neustar, and dozens of smaller vendors would disagree. And even Facebook and Google offer tools for marketers to assess the effectiveness of their customers’ spend.

But it is true that Adobe has massive scale.

The company’s Experience Cloud suite of marketing, commerce, and analytics tools takes in trillions of data points from thousands of sources, including 80 of the 100 largest retailers in the U.S.

So there’s definitely significant data feeding the company’s technology.

There is another challenge to media mix modeling (MMM) however: it takes significant time and scale. Because MMM needs to account for historical variance, seasonal fluctuation, and external drivers of customer behavior, the more scale you have and the more time it’s running, the better its insights will be. Existing Adobe customers who already have data in Experience Cloud will get value in just two weeks, Bohra says. Brand new customers will need three to four months of ramp time to start generating valuable insights. And clearly, years of data would be better, so the system knows how to account for summer, winter, and holiday variations for your specific company and your specific products.

But the rewards beckon.

Adobe says NVIDIA boosted paid registrations 500% using its AI capabilities, which Adobe groups under the “Sensei” product name. Pitney Bowes cut over 200 order anomalies a week to just one, and Under Armour manages millions of marketing assets with Adobe’s technology.

Plus of course the major cosmetics company that cut spend 50% while increasing sales 10%.

The company started working with Adobe’s MMM solution six months ago and clearly, achieved amazing results.

It should be noted, of course, that there’s a few ways to make that happen. If any company engages in indiscriminate ad spend, it is very likely to feed bad guys operating ad fraud operations at scale. So some simple optimization is likely to cut a big chunk of fraudulent activity, and advanced technology can trim even more. Plus, massive ad spend focusing on top-funnel metrics without tracking clicks and view all the way down to actual customer acquisition and sales is also unlikely to be effective.

None of that 100% requires AI.

But AI can clearly make it easier, faster, automated, and — crucially — smarter. Plus media mix modeling is notoriously complex and easy to fail at, so doing it with a vendor that has literally all the data you need plus the smart systems to run it for you is a major bonus.

One thing that’s interesting about Adobe’s AI: the black box problem.

Or rather, the lack of it.

The AI black box problem, of course, is the frequent inability of companies to understand why their AI made a particular decision or performed a certain action, leading to challenges in knowing whether it’s actually optimizing correctly or not.

“Adobe’s AI is designed to not be black box,” says Bohra. “We want to make the marketer smarter … so they know why things are happening.”

For instance, Adobe Sensei generate purchase propensity scores across customers, looking at what is influencing a higher propensity to buy. But the result is not just a probability or a percentage: it’s a list of the top key influential factors that are building that score, such as weather, location, and purchase history.

Ultimately, there’s huge value in that.

And there’s huge value in marketing effectiveness scoring that doesn’t need to monitor and surveil people’s clicks and app installs and time spent on pages. Adobe customers do have first-party data — after all, if you order something, the company needs to know who to send it to — but does not share data between brands or track people between apps and websites owned by multiple companies.

So the impending death of the third-party cookie that marketers are worried about?

Bring it on.

“Everyone’s trying to solve what the cookie-less world will look like,” says Bohra. “We started investing in that more than five years ago.”

New Adobe customers, however, may still have to wait for these capabilities.

Despite the long testing period, marketing mix modeling is still in beta and will be generally available next year, Adobe says.

Spread the word

This post was originally published by Anonymous at Forbes (Innovation)

Related posts