Facebook and Instagram ads help online pet supplement retailer Finn Wellness LLC reach new customers. But options are limited, says Randall Stainton, director of growth.
Facebook pixels traffic off existing customers, he says. Finn turned to artificial intelligence software vendor Proxima to help identify consumers beyond its customers based who are most likely to be interested in purchasing Finn’s dog supplements. There are two categories Finn wanted to reach: interest-based and existing customer/lookalike-based audiences. Interest-based groups are dog owners. They might be interested in a specific breed, or category, such as dog food or dog toys. Lookalike-based audiences are repeat/existing Finn customers.
How does AI impact digital marketing?
Over time, the machine-learning technology identifies purchase behaviors. Finn wants to advertise to consumers likely to buy pet products. Proxima’s software maps out consumer purchasing behaviors. The algorithm takes into account the different categories consumers purchase from — consumers who also buy pet items, such as clothing or home goods. It incorporates where they live, whether West Coast, East Coast, etc., and pulls together separate audiences, which are consumers that share similar traits.
Finn runs these groups inside Facebook’s application programming interface (API). Finn can decide how much to spend on an ad and adjust the budget for the different audiences within Finn’s Facebook ad manager account. Proxima runs a “feedback loop” to track ad performance. This includes tracking where the cost of acquisition (CPA) — the fee a retailer pays for an advertisement that results in a sale — is lowest, for example. The algorithm adjusts in real-time, reconstructing, or fine-tuning, the audiences based on successful periods to better reach potential customers.
Proxima recommends merchants spend a few hundred dollars a day driving traffic to these audiences to see who responds and what traits these consumers have. By using the same advertising, the algorithm can remove some of the variables to see what resonates and what does not.
“Think of it as an intelligent game of Battleship,” says Alex Song, CEO and founder of the AI marketing software vendor.
Finn invests more of its marketing budget to attract new customers
The online brand launched in the midst of COVID-19 in Sept. 2020. At that time, Finn focused mainly on its Amazon store. In mid-2022, the online retailer wanted to shift focus to petfinn.com, Stainton says.
Currently, 60% of Finn’s overall online sales come from its Amazon store. 40% of online sales come through its website, Stainton says. A direct-to-consumer (DTC) average order value is about $43, compared with Amazon orders at $37. These haven’t shifted much since launching the brand in Sept. 2020, Stainton says. What has changed is that Finn decreased spending for customer acquisition on Amazon and, instead, is investing more in paid social marketing for its DTC website.
In January 2022, 80% of Finn’s total advertising budget went toward promoting on Amazon and 20% toward promoting the DTC website. In January 2023, they reversed the ratio, Stainton says.
“The reason we’ve been able to make that switch is because Proxima allowed us to expand the breadth of our ad spend on Facebook and Instagram,” he says.
Urban Millennials and Midwestern housewives
Finn’s demographic of shoppers are primarily female, ages 35-55. There are two behavioral, or audience, types. There is what Stainton refers to as the Urban Millennial, who are typically women buying for their first dog and live either on the West or East Coast of the U.S.
The other type are Midwestern housewives. These are young couples where the female typically makes the majority of household purchases. This region spans from Minnesota to Texas.
Proxima allowed Finn to split groups into a younger Millennial bucket of 35-45-year-old shoppers and the other into the housewife bucket of 45-55 year old female shoppers.
Without Proxima, Stainton could have built and tested different buckets of customers over time, he says. And once he had a big enough record of each, Stainton could create and start segmenting advertising budget toward each bucket, he says.
Instead, the software allows him to create the buckets now.
“We don’t need to have 100,000 customers to create a good audience because Proxima is identifying those customers for us and helping us to create those audiences,” he says. “We have an approximate audience for Millennials and one for housewives that differs from what we get from Facebook.”
Finn uses AI to lower advertising costs
Finn devotes about 30% of its advertising budget to its Proxima campaigns. Of that, Finn allocates the majority (80%) for marketing toward interest-based audiences who have yet to shop at Finn. 20% is devoted to lookalike customers.
“Interest-based audiences are so much broader,” he says. “There’s a lot more dog owners than there are [Finn] lookalike customers of ours.”
Finn’s shift in strategy appears to be working. Over time, ad costs have decreased, Stainton says. Finn’s CPAs have dropped significantly to $42 in December 2022 from $72 in March 2022. Stainton attributes this decrease to the brand’s overall advertising efforts, including using Proxima software.
Where does the data come from?
In early 2021, Apple’s iOS 14 changes gave consumers the option to opt out of online activity tracking. Facebook parent, Meta, lost visibility for a large portion of consumers. It could no longer tell when someone clicked on an ad that converted into a sale. The social media company told investors in February 2022 that it estimated the iOS change resulted in the loss of about $10 billion in sales.
Proxima has data from more than 15,000 businesses and retailers that have shared customer database information with one another. Smaller brands pooled their customer data together. Retailers also provided Proxima access to their email service and SMS providers.
“When I barter my audience for yours, I don’t pay anything to anybody,” Proxima’s Song says. “I’m trading something I already own.”
Merchants also provide encrypted transactional information from a retailer’s website store through the payment processors (e.g. Stripe Inc., PayPal Holdings Inc., Venmo, among others). The information is encrypted, “anonymous,” Song says. It is coded so it can be recognized when shared with Meta, TikTok, or other social media platforms, allowing Proxima’s algorithm to track progress.
What it costs Finn to use AI
Finn maintains a yearly contract with Proxima and declined to share what it pays for the service.
The online retailer runs its website using Shopify. Stainton says he is interested in trying Shopify’s software, which also uses AI to test different audiences. Shopify’s plugin price starts at about $2,000 a month but is currently only available to Shopify Plus merchants.
Stainton says Finn is not ready to upgrade to Shopify Plus yet, but plans to do so in the next couple of months.
“The gains that we see off those campaigns have to cover the extra percentage we’re paying Proxima and the base fee — which they do,” Stainton says, without revealing how much they pay.
Social media strategy in 2023
Moving forward, Stainton says the brand intends to expand its reach on social media in 2023. For example, the online retailer has learned that videos that perform well on TikTok almost always perform well on Facebook, “but not the other way around,” he says.
So, once something performs well on TikTok, Finn adds it to Facebook and begins tracking results with Proxima.
“When we get a winner there [TikTok], we dump it straight into Proxima at scale, instead of dumping it in and seeing how it performs, [paying] more day over day and spending two weeks to scale it,” Stainton says. “We can drop it in and have a high degree of confidence that it’s going to perform in those audiences.”
Stay on top of the latest developments in the ecommerce industry. Sign up for a complimentary subscription to Digital Commerce 360 Retail News.
Follow us on LinkedIn, Twitter and Facebook. Be the first to know when Digital Commerce 360 publishes news content.Favorite