Claus Lauter: Welcome to the eCommerce Coffee Break Podcast. In today's episode, we discuss the challenges that retailers face today with paid media and how you can adopt AI technologies to successfully manage paid media. Joining me on the show today is Drew Smith, co founder at upp.ai.. So let's get started.
Voice Over: This is the e commerce coffee break, a top rated Shopify growth podcast, dedicated to Shopify merchants and business owners looking to grow their online stores, learn how to survive in the fast changing e commerce world with your host Klaus Lauter and get marketing advice you can't find on Google.
Welcome to the show.
Claus Lauter: Hello and welcome to another episode of the e commerce coffee break podcast. Today we want to talk about how you can make paid media pay. Now paid media was a bit of a pain in the neck in the past because a lot of manual things you need to do, not things you need to learn and algorithms changing all the time.
Now with AI around for quite a while now, things have changed quite a bit and we want to dive into this topic and with me on the show, I have Drew Smith. This is the co founder of upp.ai.. And he has a big background when it comes there and they're helping a lot of merchants with their strategy when it comes to paid marketing.
So I want to welcome to the show. Hi, Drew. How are you today?
Drew Smith: Yeah. Thank you for having me. Really excited to be here. Yeah. Look forward to the chat.
Claus Lauter: Drew, what are the biggest challenges retailers face today when it comes to managing paid media campaigns?
Drew Smith: Oh, that's a big topic. Um, I think there's quite a lot of uh, quite a few challenges I think.
I mean if you look at it from their interaction from the consumer side, first of all you've got more channels than ever. So you've got platforms like Google and you've got Bing and then you've got social media platforms like Meta and Instagram and TikTok, and then you've got marketplaces like Amazon, um, and they're all competing for the audience and they're all competing against one another.
They're all evolving their own technologies. They're all advancing their own AI. Um, that means that for paid media managers, they've got, uh, you know, definitely an extended buying cycle where they're trying to advertise To the same audiences in different locations across different devices, um, through different media interactions.
Um, you then also got for the paid media manager themselves, they've got to constantly be on the, uh, ahead of the curve around the adoptions of technologies and the approaches and the best practices. And then, you know, as always, um, the, I think the biggest challenge that, um, paid media's managed to have today is everything is becoming, uh, paid media is becoming such an important and integral part of a retailer.
You know, it's driving typically anywhere between 40 to 70 percent of the total traffic to a website that, the media spend has become so important that it's really important that paid media managers align their business requirements and add to their media goals and objectives and that's actually one of the biggest challenges that we often see is there's so much to learn and there's so much in the process and the kind of treading water scenario that actually By ensuring that the strategy and the business objectives are deployed successfully is often is often sometimes overlooked or just not, there's not enough frequency of adapting in that environment.
Claus Lauter: Makes perfect sense. Now, as I mentioned, there's a ton of different platforms. Today we want to talk about the one that is around for the longest time. It's Google. Google is probably also the biggest player in the game. And they came relatively early out with, um, performance mics, I think about two years ago, if I'm not completely wrong.
And we're relatively early to the table when it comes to AI. Let's talk about a little bit more about that. Um, what has this shift towards AI powered internet advertising, um, done to the landscape for e commerce businesses?
Drew Smith: It's done a huge amount in terms of the evolution of AI and how people have kind of adopted it.
If you go back, you know, only only four years ago, it was still quite have a kind of analog manual experience for the PPC managers. And you had to kind of really have a depth of knowledge and specialties around kind of like you're, you're doing bids at a manual level. It was highly complex, highly configurable rules based programs.
And then, you know, 2019, they released smart shopping, which was the first kind of walk into reinforcement machine learning. Um, and I remember that time there was a lot of. Uh, obviously talk in the market around it and, you know, with AI and automation, there often means things like kind of black box scenarios where, hey, media managers weren't becoming aware of things like, you know, they lost transparency around where products, adverts were being placed in terms of devices, they were getting kind of limited access to understanding of, audience insights and how these automation programs were deciding things around things like bidding.
But what we did see pretty much 12 months later in kind of 2020 that most paid media teams and most retailers had adopted smart shopping. And we've seen a similar thing with performance max that at first there is often quite a Concern, um, people kind of jump in, jump out of using the technologies. Um, but what you can kind of see again is Google has delivered a product that actually ultimately outperforms the old ways of working, um, provides a more simpler and sophisticated kind of tool.
Um, although there are the drawbacks that I think some of the challenges that pay media managers have is they kind of, you know, you've really gone from a kind of rules based content world to a mathematical problem, um, where you It's a real kind of data science problem, and that's quite a gap in kind of learning and expertise a paid media manager has to go through.
I think there is very limited resource in the market today to really teach and educate paid media managers of how to work with a machine learning driven product. Most of these, the background of a paid media manager isn't from an engineering or data science background, and they've kind of been thrust upon with these technologies.
As I say. Although actually fantastic, and actually if you look at the end result, most people are really happy with what Google have produced. But there definitely is the fear around how do I control it? How do I make the most out of it? What could I be doing? Um, and there is very little education around these technologies, which I think is a challenge, um, for them and a lot of the kind of old ways of working have pretty much been fully automated as well.
Claus Lauter: I have been with Google ads for, I don't know, 20 years and I think most of my white hair are coming from that. And as you said, it was rule based. Um, so a complete shift there. And as you mentioned, not everyone is a data scientist. So tell me a little bit more how performance makes campaign work and why are they sometimes hard to manage?
Drew Smith: Sure. So I think the, you know, if you, if you, if you go back in time, paid media managers, you know, especially, especially a retailer, you've got thousands of products, uh, products in different states. You've got new products being launched, you've got promotions going on, you've got products, you might be end of life thing, you've got competitors.
Um, paid media managers we're able to really control things quite concisely, like right down to the product level. They could choose exactly what products they were bidding on and the level of bid and the, the volume of the bid and exactly how much spend was going on. Um, and with all that control though, it became very, very complex and just if you just think about the mathematical problem around, you know, you've got 2000 items to sell all different states, it's a really easy reason to understand why Google said we can provide a solution that automates all these kind of laborious mathematical tasks.
I think, I think, you know, for the long term, it's a, it's a, it's a great solution, I think. The challenge with that, though, is that, yes, it's become quite black box and how do I do the right thing when I use a Performance Max solution? So, for anybody that doesn't know, Performance Max is a solution that for the first time actually from Google, does the, uh, works across the whole of the marketing funnel.
So it goes all the way from your top of the funnel with display advertising and brand awareness. Right down to the kind of shopping campaign performance Process it heavily focuses on also using the google ecosystem. So it uses youtube advertising and gmail and everything else Um, so it is kind of the campaign to rule them all um, and that does mean actually the barrier to entry if you're a kind of up and coming retailer or You're kind of deploying the first time onto the search world that still does happen It makes it a lot easier from that point of view Um, the technology that Google essentially is using is, is reinforcement learning programs.
What is, um, and that Google recommend that you put products into a campaign with, with the asset contents, display video. And the reinforcement learning will essentially work off the objectives you set in a campaign. So you'll apply a budget to it and a return of investment target, maybe a ROAS, return of advertising spend target.
And that system will essentially go and work out. How to become successful, um, reinforcement learning for anyone that doesn't know, it really works through a guest program in terms of pass and fail. So Google system will basically take that inventory within a campaign and go through testing that inventory of, does it meet the ability to spend against that return of investment?
And for the segments of content and inventory that does, it will repeat that process. And essentially we'll look to achieve the objectives that you put into the campaign. Now, while that completely pretty much works, the limitations that come with that is that the clue is in the name. With reinforcement learning, what becomes successful remains successful and is focused on, whereas other parts of your content or inventory become ignored because they failed during the testing process.
So what you see as an outcome is kind of a very typical kind of Pareto law scenario where 10 to 20 percent of the, uh, content inventory is focused on a large majority of the inventory is ignored. And we see this time and time again, when we're kind of auditing accounts, um, and you end up with basically a scenario where you kind of plateau in performance because Google becomes hyper confident on certain audiences and certain product ranges.
And it starts to ignore others. So the question to pay media manager will often go through then is how do I get more inventory active? How can I get Google to kind of relearn? And they've got, you know, they've got trading teams and leadership teams saying we need to push this promotion that's going through.
We need to, we need to, we've got, we've got a dead stock in the warehouse. We need to clear. And these challenges around business principles become very, very difficult to kind of know how to interrupt with and interact with Google's performance max campaigns.
Claus Lauter: Hey, Klaus here, just a quick one. If you like the content of this episode, sign up for our free newsletter and become a smarter Shopify merchant in just seven minutes per week.
We curate content from more than 50 sources, saving you hours of research and helping you stay on top of your e commerce game with the latest news, insights, and trends. Every Thursday in your inbox, 100 percent free, join now at newsletter. ecommercecoffeebreak. com. That is newsletter. ecommercecoffeebreak.
com. And now back to the show. So I understand. Thanks. At App. ai, you have come up with a concept that's called product data platform. Um, tell me what it is and how it basically addresses the technology. How does it address all the challenges that you just mentioned?
Drew Smith: Through our learning, and as we saw kind of Google develop these technologies, I think the really kind of simple analogy to think about it is the platform like Google is the best customer data platform in the world.
It's the best CDP. We are constantly. All of us training that, uh, those algorithms and it's getting smarter and smarter and also, you know, with everything that's going on around GDPR and, um, and consent mode, you know, audience data is becoming more and more difficult actually for most retailers to obtain, uh, from the platforms.
So we kind of see the shift that You know, the last 15, 15 years, it's all been around audience data with ad tech and marketing. It's, it's, and there's some great technology to do that. And it makes complete sense why you would put the customer at the center of the universe when making decisions. But what we could see was kind of, okay, well, the platforms are really going to own that space and they're doing a great job, but how could you impart other data sets?
And were there other data sets that were kind of being ignored? And we thought about it around, well, what about product data? Product data is fascinating because one, um, it's unique to the retailer. It's actually highly active. If you think about a retailer, you know, and a purchase cycle, you might have a customer by if you're lucky, kind of six times a year.
That's not actually many interactions with your business. But inventory and product data is constantly active. You're changing prices. You've got different stock levels, you've got sales volumes changing, and you can kind of see that activity all of the time. So that was really interesting. So the first question we actually asked ourselves is could we predict.
The probability of an item selling, and then could we calculate the cost of selling that item so we could then work out the return of investment. And by understanding that what we were able to do is actually look at kind of bringing the concepts of a customer data platform into the product world and create a product data platform.
And that's essentially what up is. It's a product data platform that works out of what's the, how often is a product going to sell? What's the cost of selling it? And what will my return be? And by understanding that what we could basically do is use all of the great technology that Google is using in performance max because it's fantastic, but also bringing this other data set that Google largely doesn't understand around not only what's going on in the media world and what's going on in the Google landscape, but also what's going on in the e commerce world and what's going on in the commerce world and by bringing inventory data and performance data alongside our customer data, uh, it's kind of a marriage made in heaven from that point of view.
Claus Lauter: Okay, that sounds for me like there is a lot of data points that you have to process to get this this match right to show the right product in the right moment to the right customer. How much of kind of a learning curve is there for your clients or basically for the system to learn the clients? SKUs, products, and the media strategy.
How long does it take?
Drew Smith: Sure. I mean, so, um, with all of these AI systems, the crucial thing is, is data. Um, and for that reason, typically a customer that fits kind of perfectly with that, they need to be spending around, um, a million pounds a year or greater on a platform like Google. Um, that's really because on the media world.
Spend kind of creates data for you. Um, and you can then really start to see patterns in the behavior. Um, in terms of actually the learning process. What's quite unique about up because of the AI world is the bigger the account, the bigger the data set, the quicker it is time to value. So an account spending greater than, uh, 2 million is, is seeing a return of investment in, in the typically in 14 days to 30 days very quickly.
And then what we see is a, is a smaller account, will take slightly longer.
Claus Lauter: Okay. That's very quick. Who's your perfect contact in the corporate world to talk to? Is it a media manager? Is it the marketing department? Is it a data scientist to whom do you usually speak to?
Drew Smith: Uh, yeah, typically we're talking to, um, chief marketing office of CMOs, uh, head of paid media teams, um, paid media managers.
Uh, also sometimes CEOs, because pay media has become such a crucial thing for most retailers, we're often talking to quite senior stakeholders within, within those businesses. And they're often to our surprise, they have a great level of knowledge. Actually the senior leadership team, it's not uncommon that a CEO will really understand the latest trends in Google and pay media.
So it, you know, it's, it's kind of typically quite. Top a top of mind and people really understand the ins and outs of it.
Claus Lauter: Yeah, I think it has to do what you mentioned before. It's, it's, it's pay to play. Um, most traffic is unfortunately paid now nowadays. So I think you need to have a mindset and a bit of experience on how that works.
Who's your perfect customer? Are there specific industries on these that you work more with than others?
Drew Smith: In terms of verticals, we pretty much work with all the verticals. So from any kind of vertical of retail, from sports to health. And fashion and electronics and home and garden. So we're kind of quite broad in that, in that sense, um, we're even kind of looking outside of physical product into things like, um, holidays and travel and that sort of thing.
Um, but to work and get really get the benefit out of up, you need a fairly large inventory. So multiple thousands of SKUs and our largest customers have multi millions of SKUs. Um, like I said before, you, you really kind of want to be spending, um, a few million pounds a year to really get the benefit of that.
Um, because what we're really saying at that point is there's so much data it's going to analyze, um, really, you're kind of at that level where the other options today is can continue to build out your paid media team, um, and add further and further heads, um, which becomes, I think the challenge again in the market is.
expertise are hard to come by. Um, and also to keep these people in house is quite challenging. Alternatively, you could go to an agency. But again, sometimes the challenges really depends on how good that team is within the agency. So you kind of have the option of adding people to the problem, or some of the more laborious tasks and day to day, that's exactly what up is here to do is kind of allow up to deploy your media spend.
And what we've actually found and really tried to do with a paid media managers who are kind of our main user. Is actually help them move into a world of strategy if they can focus on actually ensuring that the strategy is defined to make sure that their business objectives are deployed correctly onto platforms like Google, and they can spend more time doing that and doing that analysis.
Up can take care of, you know, managing the campaigns and managing the media spend. I
Claus Lauter: think it's much more fun to do strategy than to sit in front of your ad account manager software and staring at numbers the whole day. So I think it's a perfect match for AI. Could you share some kind of success stories or case studies?
And you don't need to name brands that are of companies that work with you and what kind of results they saw.
Drew Smith: Yeah. So we look, we work with some key kind of brands such as, um, Charles to it, uh, there've been a worse number of years and they've seen amazing results. We also work with, uh, as an example, one of the major DIY home retailers in the UK, um, they're a great example from that point of view, you have.
Since deploying up, they've been able to really focus on strategy and what they are, are they trying to achieve out of the channel? Um, from their point of view, why did they join up? Again, it's around that problem that we saw that they were spending 20 plus million a year with, um, but actually only around 3 percent of their inventory was active prior to joining up.
Since then and running up for a Uh, for nine months now, they've seen product visibility go for a go up to around 84%. They can see that, um, year on year revenues have increased by about 28%. And, and, uh, actually in performance max, uh, we can see that spend has been able to be increased by 30%. So they're able to spend more.
a higher efficiency and more inventories active, which is kind of a perfect combination for them.
Claus Lauter: Yeah. That's a perfect and very sound outcome there. What's a typical onboarding process for a new customer client of yours? What kind of steps are involved there?
Drew Smith: Yeah. Um, I think with most technology and knowing kind of how it feels to be a retailer, the biggest fear is always kind of onboarding and the time and the prep it takes and Whether or not this is an IT project.
So, uh, from, from the early days, we always made the decision that we have to make it really, really simple and an enjoyable experience to onboard with that. So actually the ecosystem is already there. So we, we actually try to make it quite plug and play. So to integrate most clients will typically take five to 10 business days to actually go live with that.
We integrate across the Google, uh, estate. So Google ads and Google merchant center and all the analytics platforms, uh, as a three 16 GA four. Um, and then also because up focused on the kind of e commerce and commerce world, we also integrate into into web platforms such as, uh, Magento and BigCommerce and Shopify, uh, and ERP systems like NetSuite.
Um, we have already pre built configurations, so it's actually pretty quick. Um, and then we have as typical kind of public APIs for other usage. But as I say, most clients will go live within five to 10 working days. So it's quite a light implementation.
Claus Lauter: Okay, that sounds good. Tell me a little bit about your pricing structure.
How do you charge? What kind of range is that?
Drew Smith: So it's pretty straightforward. We actually charge a percentage of media. Um, and that's typically, um, based on, um, the size or size of the media spend. So essentially, um, The larger you are, the lower the percentage and as that customer grows, that will change.
Um, we also try to make sure that we work from a seasonal point of view as well. So, uh, we work on a typically on a month in a year program where peak season, Q4, a lot of retailers are spending a lot more, maybe in summer months you might be quieter. So we find that way works well. And also the whole idea with up is that our customer success is our success.
So we really do try to align with the customer around our, even the way we price to make sure that we're, we're here to make sure the customer grows successfully. And we, and that's how we also charge on that basis as well.
Claus Lauter: I wonder as we come to the end, I want to ask you about what do you see coming up for the next 12?
18, 24 months. Will AI completely take over or will be there still be a human component within the whole marketing?
Drew Smith: No, I don't think we're at a point where AI is going to fully take over and when it's not Skynet yet, I think, I think we're all going to be, it'd be fine. Um, but I think, I think, I think it's really exciting actually.
I think. I think that famous quote that, uh, for paid media, especially paid media managers, it's the ones that, um, you know, you won't lose your job to AI, but a paid media manager that knows how to use AI effectively, that's how you could lose your job if you're kind of agnostic to that and kind of refuse it.
The paid media managers that I get to work with, Where they're constantly curious about the next evolution. They're really excited what they can see from Google with things like the Gemini project that they're working on. They really get why Up exists and the time saving and the performance improvements.
Um, those are the exciting things. I think what it allows People do is focus on actually high level, higher value, um, projects, you know, that there are, and that's always been the way of technology, right? Um, it's being able to save time and improve a proof of process and a customer experience, but for the pay medium managers that we work with, it enables them to actually do much higher value work, get work done, uh, explore other opportunities and other channels.
You know, most customers that we, we speak to, they may be. kind of on two to three paid media channels, but they kind of typically have one single strategy. One channel will be leading the way by a long way. Removing some of the laborious tasks and the more kind of day to day tasks allows them to start to really pick up on the other opportunities that they have in market that prior to that, they just couldn't get to.
Claus Lauter: Okay. Before we come to the end of the coffee break today, is there anything that you want to share with our listeners that we haven't covered yet?
Drew Smith: I think the exciting things from Upsworld as well is that we are going multi channel. So as we see, uh, customer consumer journeys change and evolve over time with AI development and with, um, you know, the battle that's out being played out across TikTok and Meta and Google, um, the world's becoming more and more complex and more diverse in terms of the opportunities that are there.
Our retailer, um, take, take control of those opportunities. So for up, we've historically always been kind of Google focus, but we're also launching into Microsoft and Meta, uh, and we believe there isn't a great solution out there at the moment really helps product advertising effectively. Um, and so we really want to be the, the, the one stop shop for a pay media manager to put all of their product advertising through.
And make it a really simple, transparent, um, uh, experience for them so they can maximize all of the opportunities across their estates.
Claus Lauter: Sounds good. Where can people find out more about you guys?
Drew Smith: Yep. So the website is www. up. ai. Um, that's our website that has all of our context on there. We also run a, um, algorithmic media health reports that customers can fill out, and that really helps, um, customers understand where they are in terms of AI today and how they're, how they're set up and how their media is set up.
Um, it's, it's an educational tool. Kind of works like, um, evaluating, um, your credit score and it will give you a good understanding of actually how, how you're currently set up in your current ways of working. Um, we find that a lot of our interactions at that point really start to help educate what the best practices are, where, where today is and what, where people need to focus and improve their accounts.
That's free to use on our website and people log in.
Claus Lauter: Okay, perfect. I will put the link in the show notes, then you will be just one click away. Drew, thanks so much for giving an overview what AI can do for media managers out there. I think it's a massive step forward. Um, Google has the biggest reach out there and I think, um, everyone who has a good budget and is looking into AI should just check you out and see if there's a match.
Thanks so much for your time today.
Drew Smith: Thank you very much.
Claus Lauter: Hey, Klaus here. Thank you for joining me on another episode of the e commerce Coffee Break podcast. Before you go, I'd like to ask two things from you. First, please help me with the algorithm so I can bring more impactful guests on the show. It will make it also easier for others to discover the podcast.
Simply like, comment and subscribe in the app you're using. using to listen to the podcast and even better if you could leave a rating. And finally, sign up for our free newsletter and become a smarter Shopify merchant in just seven minutes per week. We create content from more than 50 sources, saving you hours of research and helping you stay on top of your e commerce game with the latest news, insights, and trends.
Every Thursday in your inbox, 100 percent free. Join now at newsletter. ecommercecoffeebreak. com. That is newsletter. ecommercecoffeebreak. com. Thanks again, and I'll catch you in the next episode. Have a good one.
