Bill Michels, General Manager, Retail Media Platform at Moloco

Cathy McCabe:
Hello, and welcome to another episode of the RETHINK Retail podcast. I am your host today, Cathy McCabe, and today I’m going to be speaking with my guest, the wonderful Bill Michels. I’m going to tell you a little bit about Bill today. Bill’s responsible for Moloco’s retail media platform where he currently oversees both the product and the business development functions. Bill brings more than two decades of successful leadership across product management, data strategy and business development at some of the most definitive companies in advertising and such. Previously EVP of product at The Trade Desk where he touched multiple product areas from identity to CTV.

Before that, he was the chief data officer at Foursquare after it merged with Factual, where he was the chief operating officer responsible for product engineering and data partnerships. So lots of accolades there, and also having worked as a senior director of product management at Yahoo! working on search where he launched and led Yahoo! BOSS and international search monetization, and also working at UBS in equity research covering both technology and telecom. Bill holds an MBA from Columbia Business School and a BA from Colby College, and he lives in Southern California and works really closely with the Moloco retail media business and product teams across the globe. Amazing. Welcome to the show, bill.

Bill Michels:
Wow. Thanks so much, Cathy. Thanks for the nice intro. It’s longer than I thought, but it was great to hear it and I appreciate the nice accolades there. It was nice of you to say.

Cathy McCabe:
Yeah, an outstanding career path, I think. What’s been your biggest highlight, do you think?

Bill Michels:
Oh geez. Biggest highlights? I don’t know. I think I always have enjoyed sort of launching and working on building out new businesses so I draw on those. Right now, I think we’re doing some… I’m really enjoying a highlight of building out the retail media platform at Moloco. We’re seeing great traction and great progress there for all of our customers. Yahoo! was a great one too. We built out a whole new search set of APIs. That was fun. Factual was different. It was also really fun. We went from zero to a large business around a large data business. That was definitely a highlight and the first time I sort of started from no product to actually building, or no… Under 10 person company to building it to a much larger one. Then, yeah, Trade Desk’s a special place too. They’ve got a fantastic product, fantastic business and I learned a ton there working with my colleagues there. So I don’t know. I touched all of them, but I think the theme is it’s always nice to see something you work on come to fruition and to make customers happy and to build a healthy business.

Cathy McCabe:
Yeah, amazing. Amazing. I guess obviously one of the key topics at the moment is what are your thoughts around the current state of retail? What kind of trends did you see in 2022 and what do you think that’s going to come to us in 2023?

Bill Michels:
Good question. Let me come at it, I guess… Cathy, I come at this from the ad side so I follow some of the macro trends in retail just in general, but I’m going to be able to speak to more on the advertising side. So yes, retail… On the macro side, I think we’re seeing this sort of slight pullback from the COVID surge, which especially on e-commerce, but every data point I’ve seen, which is pretty fascinating, is that really if you just sort of… There was this big bump for COVID and then a leveling, but if you just look at it, it’s still a very healthy growth year over year.

All the businesses in aggregate themselves are doing incredibly well even though they may be slightly flat, especially the e-commerce ones, year over year. The way I think of that is the trend towards online and through mobile apps is continues. I would think also sort of how these customers acquire users and new customers is constantly evolving and changing. They’ve navigated through a lot of changes on how mobile privacy works and how they leverage those advertising tools, whether it’s Meta and Google and on the open internet, to effectively grow their business is always changing. I think ’23 will be another level change of that. I think retail media plays a big part in that.

Let’s see. I think obviously there’s macro stuff going on, so interest rates continue to be rising around the world. That plays into how people consider their spending. I don’t know if… It depends on the type of retailer how subject to that you are, whether it’s discretionary or must spending, but each of them are going to have to navigate through how consumers consider their own personal P&L and how that impacts their business. But also it also plays into how they want to spend their advertising dollars to get the most return on it based on how their consumers are acting. I guess those are some high level points, and it’s all these different aspects of the economy and the products are all playing out in realtime all the time, so it’s great. It’s fascinating to see them constantly changing and kind of see how the earnings come in each quarter and what the different trends are.

Cathy McCabe:
Do you see much difference across the globe in terms of between Europe and APAC or the US, or is it pretty much the same?

Bill Michels:
There are differences. I would say, okay, for… I’ll speak to it from the advertising perspective. There are multiple… A lot of retailers are seeing how much money and how much profit can be made from having a healthy ads business on their properties. Amazon is probably the leading example there. I think they finished the year close to… It was 38 billion, so getting really close to a $40 billion ads business. That’s way more profitable than their existing retail business, or AWS business even for that matter. Other retailers are taking notice and they’re saying, “Oh, I have a high volume, low margin business. How do I supplement that P&L with a very rich, high margin revenue stream?”

So they’re looking at ways to do this but at the same time not hurt their GMV, to not hurt their consumer experience. They don’t want to jeopardize their course, they’re looking for intelligent ways to layer it in. That trend continues. To your question of how it differs, I would say in the US there’s probably maybe a year or so ahead where we saw last year lots of launches of multiple retail media networks as advertisers launched these and tried to figure out how they can build businesses around them. Europe is not far behind. We’re seeing those launch there too, and in APAC we’re seeing that coming as well. So maybe it’s just a staggering of how we’re seeing these launches, and then we’ll start… Over time we’ll probably start to see how the products they build are slightly different too.

Cathy McCabe:
Okay. Interesting. I know obviously you’re… In terms of building, retail media platforms is your primary focus at Moloco so can you talk a little bit about what that means, what they are and how they can benefit?

Bill Michels:
Yeah. Yeah, exactly. What we do is… I’ll go back to… Every commerce platform, it could be a retailer but it could be other types of commerce, it could be food delivery, could be travel platforms. Anywhere there is a group of… You’ve attracted consumers to look for goods and services in a platform and that’s provided by a set of merchants. What we do is we help those companies launch a performance based automated ads product inside their platform. If you have users, if you have a merchant base and you want to include an ads business in there, we give you a ready to go business that you can start selling ads that are what we think is very performant and very relevant to your users that give you a durable, high margin revenue stream to include.

I’ll give you an example. Right now we’ll work with platforms. We can see anywhere up to 5% of their GMV you can earn in ads revenue. That ads revenue can… 75% or higher gross margin. We’re seeing companies launch these ad revenue streams with just a couple headcount on them and collecting revenue, like I said up to 5% of your GMV, so it can be really effective means to supplement your core business. Then there’s a couple models on how to do it, but what Moloco does is we’ll give you the tools and the machine learning to make sure that those ads are relevant and performant.

What we mean by that is we want to show ads in front of your users that are highly predictive of what we think they’re going to click on and what they’re going to buy so it helps… It’s a good user experience is one, so that you’re seeing things that are not jarring, they’re seeing things that are great suggestions, you’re able to pull in sort of item discovery, skew discovery there, but also it’s great for the advertiser. It’s the advertisers spending and they’re seeing really high return. We see an average… On the live platforms, we have now 800+% return on ad spend on what they… For those merchants.

I think of it as a three leg stool there. You have the commerce platform itself, it’s getting a super high margin revenue stream. You have the merchants that are advertising in that platform. They’re getting high return on that ad spend and that 800%. Then there’s the user there too that’s discovering new products, that’s having a relevant user experience inside the platform. If done right, now you can balance all three of those and really have a durable business line there. It’s great. It’s a big trend. Obviously it’s been going on in the US now for a little bit longer, but it’s expanding and it’s also evolving where it’s becoming more performance based and less… I would call it brand awareness based. I can go into those distinctions a little bit too later.

Cathy McCabe:
Yeah. Okay. That’s really interesting. I guess how… Do you find it challenging to sort of encourage retail brands to experiment with this or is it something that they’re quite open to? How does it compare with sort of more traditional marketplaces?

Bill Michels:
Yeah, good question. We think of it as there’s one divide, which is first party retailers and third party retailers, so first party being you hold inventory, third party being your more of a marketplace, you don’t hold the inventory, it’s a set of merchants. We work with both. The distinction, though, we are seeing better success with the third party because they have that captive set of advertisers already. Those suppliers or the merchants that are giving them those products are in a merchant portal, they go to that site and they manage their sales and they see what they’re earning through that platform.

Then when you give them a tool inside that portal to go start advertising or promoting their products, it’s very sort of organic and native to that experience so you have this captive set of audience and right away you can turn on that ad spend right away. They can track inside of it the return they’re getting and you don’t need as much of a, or any really, dedicated sales team to go find them. If you’re buying through a wholesaler and you’re doing wholesale, then you do need a sales team to go out and say, “Okay, great. I don’t know, you may not know this but your product sells in my store. You may want to buy banner ads. We can give you data and metrics around how that return is you’re getting inside of it.” But you don’t have as much of that organic relationship or existing so it’s a lot easier to turn on for what I’d call these marketplace commerce platforms.

Cathy McCabe:
Sure. Okay. Okay, interesting. You mentioned earlier in terms of how technology’s evolving and things like AI and machine learning and how does that make… Can you give some examples of how that is really making it easier to drive sort of personalization?

Bill Michels:
Yeah. Yeah. Yeah, good question. I’ll try to frame it this way. In general, you try not to focus on the technology. You try to focus on what is the problem you’re solving, what is the solution you’re providing to somebody. We solve these problems through machine learning and generative AI, but what it is is we’re making predictions around what is the right product to show to the right person in that commerce marketplace or platform. Where we leverage it heavily is if you’re in, let’s say, a food delivery app or a retail app, we’re going to show you suggested products in the form of ads but it has the product name and the detail. It’s called out as an ad but it does have… It’s more than just a generic banner. It’s got details of the product. We’re going to give you a prediction of what we think you’re most likely to engage with, most likely to convert to and buy from.

That’s based off of every event we’ve seen in that platform. Our models would then train off of every browse, search, add to cart and purchase. We see all those in realtime and we update that model multiple times throughout the day as all these training events come in. It just makes us… It allows us to put that much more of a relevant prediction in front of you. It’s subtle but you try to think of this as early on, what you could do is someone could buy a banner and put it up there and say, “Here are diapers. Okay, I hope people like this. I know…” Maybe they have some sense of like, “I’m going to show this to a group of people I believe to be new moms.” That’s a good… That’s better than nothing. It’s a good, smart tactic. “I’ll put that up there and I’ll charge a CPM.”

For us, it is… We’re not sort of putting them in segments or relying on other data elements. We’re making really impression level for that user predictions of what we think they would most likely want to purchase next. To do that at scale, you have to really invest in a lot of machine learning operations and so it’s… Think of it as AI. That is… We run a business off this now at very large scale. That happens multiple times every second. There’s not many, I would call it… It’s fun to play with ChatGPT and -3 and these are really amazing tools, but we actually have them in production running businesses right now too. I don’t know if there’s that many, actually. This runs off of what we call a transformer model, which is the same one that that’s in ChatGPT. That is… We have one in production that runs our retail media platform business.

It just happens that we think that that is the perfect model for making predictions based on image and based on a time series of what we think someone is going to do next. How that plays out for the advertiser too is it gives them really strong returns. 800% ROAS is fantastic if you can… Sort of a black box to make money. You’ll spend more if you can get that return on each ad dollar. So it’s great for them, and then we see that as the, for the retailer commerce platform, the best way to build a durable, long-lasting high margin ads business. It’s less dependent on sort of brand budgeting, I’ll go back to that diaper example, or maybe it’s putting banners up. Those are useful, but the business around immediate ROAS or sort of immediate seeing that return is the more durable version. We see the commerce platforms that implement that are doing better over time.

Cathy McCabe:
Mm-hmm. Amazing. I mean, it’s super fascinating and I think now it’s kind of… As it continues to evolve, it becomes… It’s making it sort of really targeted, really targeted, really relevant, and that’s what really resonates with the end consumer.

Bill Michels:
Yeah. Yeah. Exactly. It’s funny, we’re trying to pick our vocabulary of how we describe this well and one thought is maybe we stop saying targeted because there is no targeting, in that a human didn’t select or someone didn’t say, “Let’s target males 35 to this that have bought these products like this in the past.” That’s targeting. This is just a guess. It’s a very educated predictive guess that goes like, “Hey, based on all the events I’ve ever seen of everybody in here, including this person and others that have been updated in the last second, I think that this has the highest likelihood to please this person, to put them into… To convert them.” It’s a prediction more than a targeting. At the end, they still want the right result, but it’s a shift in that framework of how it’s built and how it’s scaled. It’s probably… It’s closer to Google and Facebook have this of how they provide buying for commerce on their platforms. It’s predictive that way. It’s for retailers, which is fantastic new inventory to do this buying on. We want to give them those same tools that you would use to be buying through Google and Facebook and a select few others.

Cathy McCabe:
Yes. I mean, it becomes super relevant, doesn’t it? That’s what’s really key. Really key.

Bill Michels:
Yeah. You can see it in-

Cathy McCabe:
then-

Bill Michels:
I mean, that is how Amazon built their large ad business. Certainly they do brand spend, but the vast majority of it is this type of model. We also see it in a lot of other sophisticated platforms, whether it’s Instacart or DoorDash or Mercado Libre, Shopee, Alibaba. All around the world we see the same model getting implemented, but it’s something to do it well. You do need access to this technology and so they can do it, not everybody can, so it’s a valuable product because we’re seeing a lot of these different marketplaces and commerce platforms go live.

Cathy McCabe:
Amazing. Amazing. What do you think’s next for Moloco? Are there sort of other products or solutions that you’re particularly excited about, or…

Bill Michels:
Good question. What is next? Yeah, we have a very healthy what we call DSP business. That was our first product, so that is buying, helping get outcomes for customers, whether that could be an app install or using a certain amount of a game or using an app to a certain amount, a very objective based outcome buying. We’ve built a very strong healthy business around that. The RMP is a new product. We have a long, big roadmap there. We’ll be adding tons of features. We’ve just added sponsored banners and advances on how we target the ROAS, and then looking to buy offsite as well is one element of the order up there. Then we have another product called PAS, Performance Ad Server, which actually helps large… Or any size I should say. Helps apps sell performance ads themselves through their own sales teams. Those are our three product areas. They all sort of have the common backbone of around our machine learning infrastructure, and we think all of those have really great legs and to build many billion dollar businesses on top of and not only help advertisers get great returns, but also help the partners really build out strong advertising businesses of their own.

Cathy McCabe:
Okay. Amazing. Is there kind of an ideal brand or retailer that you would sort of work with or partner with?

Bill Michels:
Yeah. Good question. There’s lots of verticals and there’s lots of GOs, so what’s great about this business is it replicates everywhere around the world. Any type of retailer, I’ll start with that, so especially if they’re a 3P retailer, they have a good set of existing merchants that provide for them because that’s a captive set of advertisers. I’d also characterize other verticals as food delivery is a great one. It’s another commerce platform. The restaurants or the grocery stores are effectively are merchants too. They’re looking for returns. We’re seeing nice implementations also with travel, so where their users come and do search and discovery for different travel destinations. The hotels, for example, are fantastic merchants and they are buyers of performance advertising.

Then we’re also seeing it in digital goods. You can think of this as an app store, as an example, where people go look for goods. Yeah, you want to promote things, you need providers in there of the apps themselves or it could be online games or online tutoring companies. Anywhere that they’re selling or want to promote a good makes sense and they’re willing to pay for that performance. I’d also characterize we do better when the users are in the platform doing search and discovery. If they’re… The more users that are in there and the more time they spend, the better the models and predictions are so food delivery apps and the others, they’re great. They’re not just in and out. You don’t just SEO a page and never go back. They have a history, they know what you’re using and browsing before in the past so it’s a great signal to build these really nice predictions.

That would sort of be another example of where we do best. We shine best when there’s great user events that we can build models around, and then it’s also very privacy safe. This is all within the platform, doesn’t leave, these are all logged in experiences and it’s also… The conversion events are all right there. We see them, we can tell you exactly what your return is with high efficacy and there’s no sort of slippage or mapping across different device spaces that we need to do. It’s a very clean experience and it works well for the advertiser.

Cathy McCabe:
Great. Great. That’s amazing. If there’s a retailer or a brand out there that’s listening to this podcast, how do they get in touch? How would they-

Bill Michels:
Good question. Do we have… I guess we could just… You can go to Moloco, moloco.com, you can contact us there. You can actually just feel free to email me, bill.michels@moloco.com. We’d love to talk to you. We have a fantastic pipeline and I think we’re… Yeah, we have a lot of great case studies and examples of how we’ve helped companies quickly turn on high revenue and high margin revenue streams.

Cathy McCabe:
Okay, great. Super. Thank you very much. I’ve really enjoyed the conversation tonight.

Bill Michels:
Yeah, likewise.

Cathy McCabe:
I think I’ve learned a few things as well, so thank you.

Bill Michels:
Oh, good. Good. I’m glad.

Cathy McCabe:
Right.

Bill Michels:
Thanks so much, Cathy. Take care.

Cathy McCabe:
Thank you. Thank you, Bill.