Human + AI: The Winning Formula For Better PPC — Stephanie Richter | How AI Powers Advertising, Why Human-AI Ad Collaboration Wins, How Automation Saves Seller Time, Why AI Beats Rule-Based Systems, How Micro-Data Drives Ad Bidding (#496)

Please support the show if you liked today's episode:

• If you love the podcast, please get someone else to listen, too!
• If you enjoyed this episode of the Ecommerce Coffee Break podcast, please head over to Apple Podcasts, leave a rating, write a review, and subscribe.
• Share the podcast with your family, friends, and co-workers.
• Tag the podcast on Instagram @ecommercecoffeebreak and let me know what you like about it.
• If you like the content and would like to support the podcast, you can buy me a coffee here.
• Become a guest on the show or sponsor an episode.

00:00:00:22 - 00:00:21:01
Unknown
We have to work together. I think machine just running by machine will not be successful. And human without a machine, without an AI will probably not in every case, successful. Everyone seems to be talking about AI replacing marketeers. What's your take on that?

00:00:21:03 - 00:00:40:08
Unknown
Hello and welcome to another episode of the E-commerce Coffee Break podcast. AI is everywhere right now. Every week, there's a new tool promising to automate your marketing, to run your campaigns better than a human ever could. But is that really what's happening? Or do the best results come from humans and AI working together? To help us answer that, I'm joined by Stephanie Richter.

00:00:40:10 - 00:01:03:00
Unknown
She's the CEO and co-founder of Adspert an AI powered PPC platform that helps e-commerce brands optimize the advertising across marketplaces like Amazon, Walmart and eBay. Stephanie, great to have you on the show. Yeah, thanks for inviting me. I'm looking forward to our conversation. Let's jump right into it. Everyone seems to be talking about AI replacing marketers. What's your take on that?

00:01:03:02 - 00:01:31:13
Unknown
I think every generation is afraid of getting replaced by something. Now it's now it's an AI. I think, yes, the world is changing. I, for instance, start my career not on a computer on the I don't know what is this something before computer. Where? Where we wrote papers. Now I'm working every day on a computer. And probably when I'm ending my life, I'm not sure if I'm using still a keyboard.

00:01:31:14 - 00:01:56:17
Unknown
So everything is always in a dynamic situation. And I think it's very important that we all always be looking up what is coming next. What I'm interested in do I want to go that journey? Do I want to go in a different way? So right now it's AI. It makes my life definitely easier, but sometimes also more complicated because I get so faster answer and everything.

00:01:56:17 - 00:02:21:08
Unknown
So I have to keep up with the machine. So machine is sometimes drive me nuts. Very good point there. Now obviously we are talking about e-commerce. We will talk about marketing and everything that comes together with the podcast in the last three years. AI is a constant topic. We're talking about AI all the time now. Obviously, people are getting a little bit tired about this, and I think the human factor is actually very important.

00:02:21:10 - 00:02:44:17
Unknown
AI can't do everything. So you want to dive a little bit more into this? I humans still beat AI when it comes to advertising. Or is AI has AI completely taken over? I think it's an we have to work together. I think machine just running by machine will not be successful. And in human without a machine, without an AI will probably not in every case successful.

00:02:44:18 - 00:03:11:21
Unknown
Maybe they are really small topics where just in human can do everything by themselves and be successful. I would say human and machine or an AI has to has to work, has to work together and then they are very successful. Give an example. In my field of PPC and bidding, our algorithm is came originally from the 17th. So from the financial markets.

00:03:12:02 - 00:03:38:18
Unknown
That is where my first company where I set up my first company into it. So we did optimization for banks stock markets with algorithm from the 17th. And for us the most important thing was the machine was so slow that if we need an forecast for an hour, we have to wait several days. So but nevertheless, you want to work with a machine with an AI and we, we, we we were using AI for a long time.

00:03:38:18 - 00:04:11:13
Unknown
Then we moved into digital marketing advertising. So we use AI for a really long time, but where we need humans and still that was 20 years ago is for making decisions, strategic decision transfer ideas into a language that an AI can produce. The optimal, the optimal work when it comes to PPC on my end, for instance, is I need somebody in front of a computer to give the AI the right instruction.

00:04:11:13 - 00:04:32:09
Unknown
Everybody is now calling it prompt. In the past it was a strategy into the right, a cost, terms on the right Roas. You have to make a decision on which product I want to have a certain goal and that is a strategy. So it's a translation from a strategy that a machine can can use it. It's a very good explanation.

00:04:32:09 - 00:04:50:16
Unknown
And we want to dive a little bit more in Amazon next. When it comes to Amazon I want to talk about AI. There's a lot of data. And I think that's where I really helps all these data points that you need to have to optimize your marketing to optimize your PPC campaigns comes together with Amazon. But you say the human gives the strategy.

00:04:50:17 - 00:05:18:06
Unknown
How does that work together? So the right job for the human is to know what should the machine doing to be successful? For instance, when, when when we talk about Amazon, we have knowledge about on. Let's give you an example. On the third Tuesday in the months at 2 p.m., a woman interested in financial products searching for a cap for her partner.

00:05:18:06 - 00:05:42:03
Unknown
So that is all small data points. And for that specific case, you need a specific bit just for 2 p.m. on a third Tuesday in the months. And then and a lot of more data is coming through, is she's requesting it from a mobile phone, from which country somebody is searching for a product and all these data points has to similarity make a decision.

00:05:42:03 - 00:06:14:03
Unknown
But the human, on the other hand has to define how much can I afford in that for that? Advertisement. Can I afford €100? Can I afford 20% of my revenue? Is a lot of decision making process to give to the machine? That's the machine is at the end, bringing it into a perfect result. When I try to make it as a simple explanation, I say, imagine you entering a car and you have to make a decision where you would like to drive.

00:06:14:07 - 00:06:41:16
Unknown
Is it to New York or is it just to Manhattan, or is it to, I don't know, other side of of the US. So you have to make a decision and that is what the human has to be. But then in which car maybe, or what size of of a motor that has to define by a machine. Now obviously you have a solution that helps sellers with their PPC campaign management.

00:06:41:16 - 00:07:00:02
Unknown
I want to dive a little bit into on that how that looks in a day to day business. Obviously there might be a lot of SKUs, there might be a lot of products in a store, and everyone has every product might have a different strategy for a pain data. Walk me through a typical day of someone working with your system.

00:07:00:04 - 00:07:25:21
Unknown
So typically they checking results and if they're happy with the results, everything is fine. Sometimes they're going into our software and just giving a new strategy goal. And then the machine is adapting. If they are new to our software, then they are more joining on a really every second or third day joining the software. Just checking is everything.

00:07:26:00 - 00:07:59:20
Unknown
Is everything fine, but is there working long term with that? But they may become once a week every two weeks if they're not changing strategies or if new if everything is on a day by day and no new products coming in. So that is, I think, an advantage of our software, because we have an AI software that means you don't have to changing every day the tires, putting gas into a changing IRL that you have to do if you have a rule based system, because then you have to react on every new data point.

00:07:59:20 - 00:08:25:14
Unknown
But with an AI, you give the strategy you given goal, and the machine is responsible to delivering to that goal. So that's why it is at a certain point, if you're just running your normal business, not necessary to every day joining ads. But yeah, okay. Now that helps obviously a lot because running campaigns and managing them manually and I'm doing this for a very long time can be a little bit annoying and frustrating.

00:08:25:15 - 00:08:55:20
Unknown
Now, obviously you want to have the ear reporting back to you if something does not work within the limits. What kind of metrics or KPIs or reports does your system provide? So we have standards reporting, of course, of the basic or of the typical KPIs like conversion rate, click rate, a course, robust cost limits. But we have also an individual system as an individual alert system where people can set up their individual KPIs.

00:08:55:20 - 00:09:29:21
Unknown
They're interested also their limits if they want. If you overstep or below a certain percentage or a certain number, they can define that by themselves or give me every Monday all these steps. And if they are not in this range, I would like to get an alert. So it's quite individual. We give up to the owners to develop for everybody, the individual and and where everybody can find everybody, because then it's so overwhelming and you stare at it saying, oh my God, no, no, that's too complicated.

00:09:29:21 - 00:10:00:13
Unknown
I don't want to use that software. But also now with AI and with Claude, for instance. Or you can build your individual alerts, you can build individual dashboards. It's so much easier. I'm not sure if somebody heard an MCP. So you can connect and then build everything from scratch. I'm not even sure in the future if, say, even big dashboards needed anymore.

00:10:00:15 - 00:10:24:11
Unknown
Maybe some of us remember when Google came out and before Google you have to go through who to a lot of I want I'm interested in that. Then I go to the next part of it. You going down a tree and then Google start just with one line and you enter what you are interested in. I guess that is coming with an AI in the future as well.

00:10:24:12 - 00:10:48:02
Unknown
You just entering by voice what you're interested in and then you become your specific information back. Yeah, it's an interesting thing that you mentioned there and I'm totally with you. I'm also working with mic now, and I think everyone out there who is in e-commerce for a while and working with different apps and solutions, has 20 different dashboards to manage, and it's probably tired of it.

00:10:48:02 - 00:11:06:22
Unknown
And I think you're completely on point there that there is a better way coming. And that might be MCP or any kind solution that comes from there. Now, our listeners obviously know that AI usually needs a little bit of training time to get all the data together. How does that look with your system? It's the same year at expert.

00:11:06:22 - 00:11:38:08
Unknown
So what we typically say. So we offer a 30 day free trial. Or it depends on the size of a company, very low budget two months pay trial. And at the beginning the first two weeks we always cause a training time. So we don't recommend an performance goal in that time period. We define a cost goal based on the historical values and suggesting these kind of goals that the algorithm can learn because we building different data than what we get from Amazon.

00:11:38:08 - 00:12:02:05
Unknown
From Amazon, you get a lot of data, but we build trainings, data just for ads. But as I mentioned a few minutes before, the third Tuesday of a month, for instance, has different data then if you're collecting every Tuesday. So we combining data from Amazon. And that is our learning curve. And it depends. If you have a lot of data your learning time is shorter.

00:12:02:05 - 00:12:27:14
Unknown
If you have less data then you need a longer time. What a lot of people not considering when they're checking their Amazon account. You have a few keywords, a few Asians where you get a lot of data that you're most promising keywords, but then you have a long tail of elements where you have not so much data, and there you are, losing your money because you cannot really judge what is the best decision for these elements.

00:12:27:14 - 00:12:57:00
Unknown
And for these elements you need a little bit longer training data, but what you also need is the right combination between data. So you need to find patterns in data because you cannot spend for every single item, let's say $0.20 and collect collect five clicks. Imagine you have one of keywords or thousands of products. You are losing money and cannot make any further decision because you just collected five clicks and no conversion.

00:12:57:00 - 00:13:19:12
Unknown
What does it mean? Don't spend any money on that keyword or in that Asian anymore. Or you get one click, one conversion. Does it mean okay, spend all your money just in these single Asian somewhere in the middle is a decision. So that's why we typically have two weeks of learning curve. And it also depends if you're coming out of the season.

00:13:19:12 - 00:13:35:17
Unknown
If you're coming in season, I would love to give you a clear answer because I'm German, I like precise answer, but as you're figuring out, it's a little bit more complex. But roughly with two weeks you are in a good fight. Yeah, but we're also marketing. And marketing is a little bit of a lot of gas dimension in there.

00:13:35:17 - 00:14:07:21
Unknown
So I think it's a bit of a mix of both. Now you're not only working with Amazon, you're also working with other platforms. How does that work and what kind of data pull you from there? So how we build that spot, we are able to collect with more or less every marketplaces worldwide. So for instance, here in a in a German region, we have, for instance, eBay, we are right now getting on more or less every miracle market, miracle marketplace and apps marketplace because miracle bought us one and a half years ago.

00:14:07:22 - 00:14:38:17
Unknown
That was the next expansion phase. And then we going after a lot of new marketplaces. So right now it's Amazon, Walmart, eBay. Here we are also able to optimize Google and Microsoft. But right now we are not taking new clients on. We are fully with all our capacity. If you listen to that episode in, I don't know, 2027, you can definitely find another 50 to 60 marketplaces on our home page.

00:14:38:18 - 00:14:54:16
Unknown
Well, that's a huge advantage to have all the data in one place. So with that, you have one point of truth. Instead of having 50 different marketplaces that you need to manage individually. Can you give me an example of a brand? And you don't need to name the brand that you have worked with and what kind of results they saw?

00:14:54:18 - 00:15:21:16
Unknown
We have several use cases on our website, but one was specific in my mind and it's a typical case. They have a lot of products. They have really good structure on their accounts when they joined us, and they want to expand in other region, and they did everything manually, and we were able in two months to triple their sales just because we were able to increase and decrease.

00:15:21:17 - 00:15:55:10
Unknown
Also the long term keywords, and we give these companies a goal. We need your strategy. And they used one of our features that is not used so often. So we are able to optimize over several marketplaces. When it for instance comes to Amazon, we are able to optimize Amazon US on Amazon France, Amazon Spain, Amazon Germany. And so what they did is for all their important products, they make one single group for optimization.

00:15:55:10 - 00:16:18:22
Unknown
And then for the rest of the products is a group over several marketplaces or countries together and just give them a performance goal. And they save time. And they took all our possibilities of a machine to really optimize all them together. And then we push, for instance, more money into Spain less than Italy, because it was better or less sold that product.

00:16:18:22 - 00:16:46:14
Unknown
And we were also quite impressed by the numbers of revenue. And also at the end we we checked how many times they saved and they saved in one week over 200 hours of work. So nearly a month. But a human is not checking every keyword every day because you can also doing a day changing, you can have top of place audience.

00:16:46:14 - 00:17:09:10
Unknown
So a lot of possibilities what you can change and you should give that to a machine. Don't do it by yourself. I think that's a good example how AI can optimize the data that's just coming in. And if you double, triple your revenue, then obviously you have a luxurious problem with all the fulfillment and everything, but that's a good problem to have.

00:17:09:12 - 00:17:34:00
Unknown
Who's your perfect customer? Are there specific industries or verticals that you work more with than the others? So one goal when we set up at spot was we want to give every single e-commerce brands a chance to have a software, because often really good software is for the big brands where they have the money because it's very complex software.

00:17:34:00 - 00:18:00:03
Unknown
We have two tiers we like to work with. One is really an SMB customer who maybe can only spend $200, $100, have ten products, want to grow their business? They can use a simple version of ads where we where they get a lot of defaults in it and cannot make so much own decisions. And then we have enterprise brands.

00:18:00:05 - 00:18:24:06
Unknown
They have more times, they have individual suggestions. I want to they want to place in advert. And for them we have a more educational software where they have more choices in asphalt. We already talked a little bit about this, but I want to understand how the onboarding process works. What steps are involved? How long does it take to get started?

00:18:24:07 - 00:18:46:23
Unknown
Is there any kind of homework to do before you can get started? So I would say the homework is you need to know what goal should be reached. If I ask somebody yeah I want to spend €2,000. It's okay at the beginning, but you need a number behind it. What does it mean? Does it mean your acres could be 80?

00:18:47:05 - 00:19:31:10
Unknown
Your acres should be 20 or tens, or some kind of expectation or a number for for your business. I think you can expect that from an e-commerce company. And then you can start day and night, because we have an automatic system where you can connect your account. We are an Amazon proofed partner. Typically, we have to wait 2020 to 24 hours to get the data, an advert, and then we have a lot of tutorial videos where people can, in your own time zone, in your own time on the weekend, doing doing the week, learn how is works and then they can start everything by themselves.

00:19:31:11 - 00:19:52:20
Unknown
They can also request support from us. It depends a little bit who is interested in what kind of service. Okay, you already mentioned that you have two pricing tiers, but you also have like a 30 day trial. Talk me through that. How does that work? The same at the beginning. You have a certain days free trial. If you if you are in a higher tier.

00:19:52:23 - 00:20:14:20
Unknown
We have the alternative also to a low budget pay trial. If you want to go a little bit deeper in all the setups, for instance, we have also profit optimization possibility, but we don't offer that in the first months. You can start over, get immediately some videos how you can set up everything by yourself. And then at the end of the 30 days it stops.

00:20:14:20 - 00:20:40:07
Unknown
So you don't have to set up your credit cards, or you accidentally run into an Abu for the next or subscription for the next 12 months. We really want to have happy customers. Our customers make an active decision. I like the results. I like what Edward is doing advert supporting my business. I want to subscribe to advert and then they can choose between a monthly contract or a 12 month contract.

00:20:40:08 - 00:20:56:01
Unknown
Okay, that's very straightforward and I like the idea that you just can test this with the trial, no risk involved. And then if you have the feeling it works with your company, then you switch over into a paid plan before our coffee break comes to an end today, is there anything you want to share with our listeners that we haven't covered yet?

00:20:56:07 - 00:21:21:17
Unknown
If I'm so intensive to talk about, let's give me one last shot, something that is quite unique here. That spot is we have a scenario feature. So that means before you spend a single dollar we give you a glance. What can you expect? Of course it works. When we start our optimization that we cannot often and scenario without optimization because it's part of our algorithm.

00:21:21:17 - 00:21:46:02
Unknown
That is one thing. And the other thing is, if you don't use any kind of claw ChatGPT or at all, go off, stop immediately here, go to your mobile phone, go to computer and start working on that and using it. And if you start with just rewrite your email, just start somewhere. Start very simple. I started very simple as well.

00:21:46:04 - 00:22:12:13
Unknown
My first task was we write my email and I was impressed. What I thought if you are in your free time, asked what can I cook tomorrow? Just start with a simple task and you will see it is worth to and also and make your work life easier. I couldn't agree more. AI has taken over so much of my brain that it has become part of my work life on a day to day basis.

00:22:12:14 - 00:22:36:14
Unknown
Literally. Where can people go and find out more about you guys on our website? And we also because it is a really nice show, what you're offering here. You can also find the discount with 20% if you want to try it out and automatically use use the link and then enjoy sport and maybe you will come a happy customer.

00:22:36:15 - 00:22:50:00
Unknown
I will put the links in the show notes as always. Then you just one click away. Jeff, thanks so much for giving us an overview on how AI can help you as a seller. Making more money out of your business with working less. I hope a lot of people will reach out to you and I hope to talk to you soon.

00:22:50:00 - 00:22:51:01
Unknown
Again, thanks.