December 18, 2023
Machine Learning (ML) is reshaping advertising for the better. Speaking at Advertising Week in NYC, Nikhil Raj, VP of Retail Media at Moloco, explained how ML creates an optimal experience for advertisers, marketplaces, and customers through automation that delivers long-term value to the industry. The discussion explores how the pandemic catalyzed the need for marketplaces to build their retail media, how automation is applied to digital advertising, and why it’s critical to find experienced ML experts to provide ML-supported platforms that can generate better user experiences and results for all parties involved.
Watch the interview with Jamie Maw from AdWeek and Nikhil Raj from Moloco:
1. Why retail media is important for online commerce:
“What happened during COVID was e-commerce exploded and everybody moved from in-store to online shopping. And when they did that online shopping, you got to deliver the product last mile to the customer, and that makes e-commerce inherently less profitable than retail. So all of my clients at Bain were interested in figuring out how to plug that hole in the profit. What can we do to sort of make the P&L whole again? And the number one approach is retail media. That's where you get a pretty significant revenue and margin stream that creates that profit pool, if you will. And that was the biggest challenge around the world.”
2. Why ML is important for improving user’s experience with ads:
“Imagine if you went into your Facebook or a TikTok feed and you had something from six weeks ago or six months ago. It really doesn't matter, you're going to log off. It's the same case with retail media. When you are on a retailer website, you're going to want to see what's, or just in general e-commerce, not just media — you're going to want to see what's relevant to you right now. And so you've got to make that decision real time, and that's kind of how this thing becomes relevant for users as real time.”
“We don't really, sort of, control the algorithms. They self-learn. What it came up with is a different order of the same seven pairs of jeans for different users on the website. When it does that, somehow it figures out, based on your browsing and searching what types of images you click on, it figures out what is the right pair of jeans for you. And so it reorders the seven. We were able to double the click-through rate for this customer, and therefore the ad revenue doubled.”
3. Why advertisers need to balance user engagement and ROI with data-driven ad targeting:
“If you have a product from a certain advertiser, and maybe there are two, or three products from two, or three advertisers that are all relevant to the user, right? First we just talked about how we determine what's the right one for the user by making sure that the relevance is very high, but at the same time, you got to give the advertiser the return. How do you use your data to make that most performant for the advertiser? They have a certain budget that's looking to hit a certain return on that budget. If you show the user an ad that's not relevant, they're not going to click and buy. It reduces the revenue for everybody. For the advertiser, they don't sell that product. For the user, they sort of get into an ad blindness situation.”
“It's really important that for a given user, there's a real certain item, you got to match the two. And then when you do that, the advertiser's goals are met. For example, if I'm trying to reach a bunch of users and my goal is cost per reach and there's another advertiser who's trying to try to sell something, his goal might be target ROIs. If you have to figure out how to normalize these two goals into one, figure out what to show the user, whether this advertiser wins with the reach goal or that advertiser wins with a performance goal, and make sure that both of them are taken into account in an auction to help deliver the right ad to the user.”
Connect with us to learn more about building and growing your own ads business powered by machine learning.
Delve into the strategic importance of building an ads business for online retailers and marketplaces, highlighting its impact on user experience, seller empowerment, revenue growth, and market competition.
2024 will likely see the commerce media landscape evolve, defined by the use of first-party data and machine learning for ad optimization, internal handling of media efforts, and embracing Connected TV for full-funnel marketing.
Optimize sponsored ad performance for food delivery by tapping into the power of vertical-specific data — like delivery times and fees — to drive transactions and empower local restaurants with a risk-free, pay-per-order advertising model.
Learn how to accurately measure ad impact and leverage machine learning for targeted, relevant advertising to counteract ad blindness in retail media.