How One Ecommerce Company Used AI to Get a 3,000% Return on Ad Spend
Naomi Simson is an entrepreneur who hosts the popular Shark Tank Australia. But she was also a business owner with a problem.
“I couldn’t believe how much money we were throwing at the problem of digital marketing,” she says in a phone interview with the Marketing AI Institute.
Simson co-founded RedBalloon, an online marketplace for gifts and experiences. Since 2001, customers have bought everything from outdoor adventures to cooking classes—for themselves or as a present for someone else.
RedBalloon was spending more than $45,000 per month on retainers for several ad agencies. The company had little transparency into what these agencies were doing, but Simson did know she was paying $50 or more per customer acquisition.
“It was just not sustainable,” she says. “We were being held to ransom.”
In desperation, Simson and her team researched potential solutions to the problem. They stumbled upon artificial intelligence.
Getting More Out of Ads with AI
The biggest pain point RedBalloon had was paid advertising. Paid was a major acquisition channel, but the company wasn’t very good at it.
“We didn’t know how to do lookalike audiences or target and retarget correctly,” Simson says.
Simson and team heard about a solution called Albert that used AI to power paid media and marketing programs. Albert, which we’ve profiled in the past, analyzes data across your ad accounts and customer databases, then uses sophisticated machine learning to target, run and optimize your ad campaigns.
Simson was intrigued.
To start, her and the team ran a test of Albert over a few months.
The results were impressive.
“In the first three months, we saw actual material results,” she says.
From the start, Albert was doing things that humans were incapable of doing—or incapable of doing at scale. On day one, the platform tested 6,500 variations of a Google text ad. From there, it learned more and more from RedBalloon’s data and further optimized the company’s ad campaigns on Facebook and Google.
The results were so striking and immediate that Simson fired her ad agencies.
Perhaps inspired by the move, Albert was just getting started.
“We started out shooting for a 500% return on ad spend, which Albert achieved,” Simson says. “Now, we average 1,100%. On some campaigns, we hit 3,000%.”
Overall, Simson says the company spends 25% less on marketing now and sees at least 30% better results. In fact, the results of using Albert were so significant that Simson, ever the serial entrepreneur, started a marketing company that is now an exclusive Albert distributor in Australia and New Zealand.
But this isn’t even the most interesting part.
Albert didn’t just dramatically increase return on ad spend.
The platform found customers Simson didn’t even know she had.
Discovering New Customers
Albert optimizes your ads against a specific KPI set by humans (awareness, customer acquisition, etc.). While the system runs automatically, humans are heavily involved every step of the way. But just letting the AI do its thing can sometimes produce amazing results.
In the case of RedBalloon, Albert used the power of machine learning to find completely new audiences that were potentially interested in buying.
“I found markets in the US and UK of people traveling to Australia that I didn’t even know I had,” she says.
Turns out, Albert saw something Simson’s human colleagues didn’t: Australian expats were highly motivated to buy the company’s experiential deals, either as gifts for family back home or for themselves. Previously, the team had been advertising to people within the country.
Albert also effectively advertised to small audiences of a few dozen people that human teams just didn’t have the time to target.
“We can target micro-audiences like never before,” Simson says. “Say, men over 65 in Adelaide who love flying.”
Albert’s ability to advertise at scale makes it cost-effective to find and target all groups that might be interested in RedBalloon’s offerings.
AI Implementation Isn’t Always Easy
Simson is quick to note that piloting AI like she did isn’t always easy. And it’s not for every company.
“People look at AI like it’s a silver bullet. But you still have to do the work,” she says. RedBalloon spent months piloting Albert before fully investing in the platform.
The company also had lots of data to work with, which was crucial. AI must have data in order to perform. In RedBalloon’s case, the company had at least six months of robust advertising performance data for Albert to work with.
“If you’re not spending enough, it cannot do the work to find the right audiences,” Simson says. She notes that Albert specifically is probably best for consumer brands—and it’s not a fit for startups that haven’t yet acquired customers.
AI like Albert also requires that marketers, executives and entrepreneurs give up some control, which can be tough.
“I get why people would struggle with AI. It’s absolutely about trust. As a founder, I want to control everything, so it’s hard.”
But the rewards are worth it. And, contrary to what you might expect, AI can become a valuable member of the team, rather than a replacement for humans.
“Your team doesn’t have to be scared they’re going to lose their jobs,” Simson says. Instead, it’s more like they get an upgrade. At RedBalloon, people who were previously responsible for executing paid advertising campaigns now oversee Albert and make strategic decisions. Albert does the grunt work far better than humans, freeing up the team to invest time and resources into more valuable tasks.
In RedBalloon’s case, AI is good for the bottom line and good for people. In fact, the whole experiment with Albert has Simson asking some radical questions.
“The question I now have is: If I gave an open-ended budget to Albert, at what point do I stop growing? What if I just keep increasing the budget based on this level of return? These are unheard of conversations in marketing that I’m now actually having.”