How To Increase Margins By Moving Beyond Rule-based Pricing - Felix Hoffmann | How to Scale Profitably Today, Why Rule-Based Pricing Often Fails, What Predictive Pricing Models Deliver, Why Transaction Costs Matter Most, Why Data Quality Matters (#480)

00;00;00;00 - 00;00;20;01 Unknown Yeah, I think it's becoming more and more important to have really like a very good understanding of what are all the costs involved in a transaction or your business on each channel. The better you understand that, the better you can predict it. And if you can predict that, you can then use that to make a decision in the right direction. 00;00;20;04 - 00;00;38;00 Unknown Hello and welcome to another episode of the E-commerce Coffee Break podcast.

Are you prices driving profit or just more sales on lower margins? A lot of brands still rely on simple pricing routes or gut feeling, and that's where margins really quietly disappear. So the question is how do you set the right price for every product? You have to break this down.

00;00;38;00 - 00;00;57;03 Unknown I'm joined by Felix Hoffmann. He is the Co-Founder and CEO of 7Learnings. Felix previously led global pricing optimization at Zalando and now helps brands use it to improve pricing, profit and growth. So let's get started.

Felix, welcome to the show. Thanks for having me. Felix Many brands still use simple pricing rules. Why is that a problem?

Yeah. 00;00;57;10 - 00;01;20;11 Unknown One, it's a problem because the world is becoming more complicated and if you're using pricing rules, you probably have quite a lot of them. So there's complexity that you have to manage, especially in this world of marketplaces. There's more and more marketplaces which have different setups, different cost.

So you are going crazy with complexity and if you're only using rules. 00;01;20;14 - 00;01;39;13 Unknown And the other thing is that it's not going to be optimal. The question is always like, what words should you use? Even if you have crawled the entire Internet, what everybody's kind of doing.

Yeah. Then then what are the rules that you should be using? And I think that's the problem with web based pricing. Mm hmm.

Now, a lot of beginners in e-commerce, they just look at what the competitors do. 00;01;39;13 - 00;02;01;26 Unknown And obviously that's that's an issue. What's the problem with matching your competitor prices, like with saying, in the words of Frank Berstein. The the blind follows the blind basically as well.

And it's also the race to the bottom. You could also say that, right. So if you if everybody is doing that, especially if you're undercutting your competitor, a competitor, then you're kind of destroying the market in a way. 00;02;01;28 - 00;02;31;01 Unknown And also, you're not enough checking like like how much would you be selling if you would be more expensive?

A little bit. And then even like, I mean, nobody can match the lowest price in that market. Nobody can really afford that. And that's not really the best strategy for anybody, I think not even Amazon.

And so even Amazon or Zalando, they all use predictive pricing to understand if I am not following the lowest price, what would it mean for that specific product? 00;02;31;03 - 00;02;52;28 Unknown Does it look okay for my product? Because maybe I don't even have enough stock and that it's fine. This is what you have to analyze and what are the options outside of matching?

And yeah, that's important to check if you want to be profitable. Mm hmm. Pricing is such a complicated topic, so I'm really interested in learning more about this. 00;02;52;28 - 00;03;10;14 Unknown And you were at a London, which is obviously a massive company.

And I have a good background in this now that there's a lot of things I know you already mentioned a few is when it comes to pricing now is coming into the game. Is that a game changer when it comes to pricing or is this just another tool that you add to what you already have? 00;03;10;17 - 00;03;31;15 Unknown Yeah, I think the game changer is that that we have really unlimited compute. So there's a lot of computational possibilities which we didn't have before.

And there's also a lot of support from models which are open source developed which you can use and which you can use to make these predictions. And my opinion, what we're doing is called predictive pricing. 00;03;31;18 - 00;03;57;27 Unknown And the idea is really to predict what happens if you are matching your competitor, what happens if you are undercutting, what happens if you are maybe 5% more expensive? What happens if you're 10% more expensive?

So it's really making predictions on a very, very large scale, billions of predictions basically every day today already. And that's just possible with like the new cloud infrastructure and the hyperscalers and open source models. 00;03;57;29 - 00;04;14;14 Unknown That wasn't possible before. Mm hmm.

Let's dive a little bit deeper into that. Obviously, the at the end of the day, as a merchant, you want to make a profit. I think a lot of businesses out there presenting themselves, they're only talking about revenue and you're talking about their profit. And I think that's a bit of the wrong angle.

00;04;14;14 - 00;04;34;15 Unknown If you're running a business, what are the levers when you do your pricing strategy on really make sure that you make money at the end of the day? Yeah, I think it's becoming more and more important to have really like a very good understanding of what are all the costs involved in a transaction on your for your business on each channel. 00;04;34;18 - 00;04;56;12 Unknown Right. So there are logistic costs, there is commission costs, there's written cost out phone costs, marketing costs, and then there's especially in marketplace, it's a very difficult cost structure and the more the better you understand that, the better you can predict it.

And if you can predict it, you can then use that to make a decision in the right direction. 00;04;56;14 - 00;05;22;28 Unknown So that's the idea. And the problem is that this is true for like Amazon, for eBay, for Zalando, for about you. They all have different cost structures and you need to try to understand, okay, where should I drive my sets?

And the other thing is that you can also try to understand if I want to grow 10%. That's the other thing that that that predictive decision making brings to you. 00;05;23;00 - 00;05;48;09 Unknown If I want to grow 10%, where should I invest my money to grow 10%? Should you should you sell more on your own channels?

Would you sell more on on the land or on an auto? Where is it best to invest your discount, for example, or where is it best to invest your marketing money and that becomes more and more important because of these complexities of multichannel sales. 00;05;48;12 - 00;06;07;05 Unknown I like that you said that marketplace is everyone charges differently, different structure, sense of on so forth. It's already difficult when you're run your own store with all the costs and fees and fees and whatever is attached to get it right now, for our listeners, obviously they understand there is many of many moving data points in a business.

00;06;07;05 - 00;06;29;26 Unknown What are the main KPIs you would look for to start optimizing for? The most important data points we are using is the transaction file. So and we are looking for kind of really broken down transactions, trying to understand, okay, what for each sale you did, what were exactly the costs on that specific transaction? That sounds easy.

Everybody. You know, I talked to the CEOs in the beginning and they say I have that data, no problem. 00;06;29;26 - 00;06;48;10 Unknown But in the end, if you really ask for it, I mean, who really knows? Like what was the marketing cost, for example, on a specific transaction?

That's a very difficult question, but these are the type of questions you need to try to answer in order to optimize them. But of course you can. You can start with logistic cost. 00;06;48;13 - 00;07;11;07 Unknown Yeah, of course.

I just mentioned. And then next to the transaction cost, of course I also attribute data because yeah, we want to also learn across products for your business. We want to know how like we want to make a prediction for returns. Even if you've never sold a product as an example, or we want to understand what is the conversion of a product, even if you have never sold it.

00;07;11;10 - 00;07;32;10 Unknown And for that you also need to understand like what are similar products and that is what you need to attribute data for. So what is the product category? What is a brand name, the name of the product sizes and that kind of stuff, you know, And we also mix in, of course, data, like whether data or competitor data as well. 00;07;32;12 - 00;07;54;13 Unknown Yeah.

So it's quite a there's quite a lot of data used actually in this and this approach. And you could say in the short, the more data, the better your prediction, the more accurate your prediction and the better your prediction, the better the outcome of the optimization. Okay. I was surprised to hear better data, but in a sense it makes sense because if you're selling some seasonal products, the weather might have a huge impact on your sales.

00;07;54;15 - 00;08;27;27 Unknown Now when it comes to different marketplaces, would you recommend to have a different price for the same product on each and every different platform and marketplace? Or how does it look in real life? Yeah, I mean, if you have hundreds of marketplaces, that's probably too complicated. And so we also support aggregation of marketplaces.

So you can say, okay, I want to have the same price on marketplaces which have similar cost structures as a I mean, that itself has 16 to 27 different markets. 00;08;27;29 - 00;08;50;23 Unknown So you might not have to have a different price on each of these markets because the cost structure in terms of commissions at least is very similar. So yeah, I think the best practices really to aggregate these marketplaces so that you have similar cost structures, but for example, on your own channel, the cost structures very different and also the elasticities are very different.

00;08;50;25 - 00;09;13;07 Unknown And for example, return rate for a fashion item on Amazon is 30%. The return rate on the land was 60%. That's quite a big difference. So this is what I mean when you need to consider the differences of these channels to make sure you're making the right choices.

Mm hmm. Can you give an example of a a brand or a case study that you work with in the. 00;09;13;07 - 00;09;39;06 Unknown You don't need to name them. Just to give an example to our listeners how the implementation of a I helps was better pricing and what kind of increases you see overall.

Yeah, we work with large fashion retailers. We we typically do like three months onboarding in the beginning. You get all the data you get. We're working about one month on the prediction quality because as I said, quality is everything.

00;09;39;06 - 00;10;00;18 Unknown If you have high quality data, then you just have much, much better results. And then in the third part of the project, we are doing an every test at the moment. A lot of discussion, of course, on like what is I actually living in terms of impact. And we typically still measure that in the third part, not always, but in many cases these AB tests are also on our website available.

00;10;00;18 - 00;10;31;08 Unknown So some were like quite impressive. Like for example, we had like very, very high uplifts, I think even more than 100% uplift, something like that. You can look the exact number of outlets that for example, was just kind of 29% uplift in profit. So the uplifts are very, very high.

So the I would say the the biggest lever for everybody in ecommerce, if you have already bought your stock, is definitely pricing. 00;10;31;10 - 00;10;48;25 Unknown It's the number one lever if you want to improve your profit, it's definitely pricing because you already have your stock. So the question is like, how can I get most of the money out of that stock? And it's not to make that distinction.

It's not only short term profit, it's not only like how much do you make tomorrow or the day after. 00;10;48;25 - 00;11;07;23 Unknown That is one aspect of it, but it's also about long term profit like including the entire season. How can I make most money out of the entire season? Because it's also not making sense, of course, to sell out on a cross price, for example.

And then I've seen that. I seen that all the time. Basically, people do price matching. 00;11;07;23 - 00;11;31;15 Unknown Then the entire market at some point is sold out on a specific SKU.

And what is the benefit of that in the end Didn't make any sense that cash would have been much more useful to actually increase prices in between and just have a higher margin or the other. I think what's also happening in fashion at the moment, there's a lot of overstock in some companies and that also doesn't make sense in this case. 00;11;31;15 - 00;11;53;27 Unknown It often makes more sense actually to reduce the price early on in the season to sell of that overstock, because it's better to have lower profit in the short term than all that overstock at the end.

And these are really complicated calculations that a typical brand normally doesn't have the experts to do either. The capacity to do that even, and that's what we're helping with now. 00;11;53;28 - 00;12;15;14 Unknown Definitely. I can see the advantage of a I know some people might have concerns and say if I leave my pricing completely to a, I can go out of hand, what are the risks there?

What I did control mechanisms that you have to not have that happen. Yeah, but you can. I always love to compare as to Google maps, right. 00;12;15;15 - 00;12;37;26 Unknown So if you want to go from like I'm in Berlin today, I want to go to Munich next week actually.

And I just say, okay, I want to go to Munich, Give me an itinerary. How could I get there? Right. And and then I still check it, right?

So I actually optimizing that. It's definitely optimized and I still check kind of and then I make a choice which one I want to use. 00;12;37;26 - 00;12;55;27 Unknown And you can it's a similar thing in our case, but we made predictions for what are your options? We are then making a suggestion and say, Hey, this is what we would suggest to do, and then you can still check it, you can still review the scenario and then make it your choices and you can override the decision as well.

00;12;55;27 - 00;13;24;10 Unknown So we are basically I don't think there is a perfect price. There's just a perfect setup for what you want to achieve. And you can, for example, say, okay, I want to match all my products to the lowest market price and then you get maybe a another scenario for maximizing profit, the results will be extremely different, right? So as I said, in the short term, the power of pricing is just unimaginable.

00;13;24;11 - 00;13;42;29 Unknown Right? It's completely different numbers. Probably you have a lot of revenue if you are mentioning the lowest price. But in terms of profitability, the results are vastly different.

And I'm not saying you have to do it this way or the other way. I'm just saying like you should look at both options and then you can make a choice which one you should choose. 00;13;42;29 - 00;14;02;01 Unknown So our opinions not making kind of like turning it on and say, okay, so black box and makes my pricing not automatically it's more like you can use it to make better strategic decisions based on scenarios which we be helpful for you. I'm curious to hear from you and you don't need to give away your your company secrets secrets.

00;14;02;01 - 00;14;24;24 Unknown But obviously working on your own data is one thing, but still there's competitors on the marketplace there and there's marketplace prices. How do you train DJI to follow the same product or similar product of a competitor to find out what prices they have? Because I think it's important still not to match your competitor, but to at least know what your competitor is charging. 00;14;24;27 - 00;14;45;02 Unknown Yeah, I mean, we know that from quality, right?

So we don't call it ourselves, but we have partners who, for us, and then we are using that price as an input into our machine learning models, which then predict and know that the elasticity of demand is the most elastic around that price. So that's how we how we use it. 00;14;45;02 - 00;15;05;11 Unknown So we know that you will sell a lot more if you are matching, but we also know that you have maybe you lose money at that price. So that's how we how we do that then.

Okay. You spoke briefly about the onboarding process. So there is a bit of a training time period involved talking through it. How does it look in real life of a brand approaches you?

00;15;05;11 - 00;15;25;04 Unknown What are the steps to to get up and running? Yeah, well, that depends a bit on the data maturity, I would say, of that brand. So yeah, if you have very mature brand, they can onboard actually in one or two days I would say like brand with perfect data, let's say that. So yeah, our data requirements are really public.

00;15;25;04 - 00;15;47;27 Unknown You can download the requirements and you can see that as I mentioned, typically brands think that they have very good data, but then when they start to work with us, then they realize there's some problems in their data and we typically in the onboarding phase have to fix that data quality issues for us and to increase the forecast accuracy. 00;15;47;29 - 00;16;19;06 Unknown So that's again takes typically 5 to 6 weeks, I would say. But there's also occasions where we were much faster.

So there's basically a quality gauge in the beginning where we actually we have hundreds of tests running who check your data quality and we make sure that it's really perfect when gets into the prediction stage. Yeah, and then pretty early on after six weeks, you get access to your front end where you can run optimizations and then you can start using that, using the tool and also customizing the scenarios which are interesting for you. 00;16;19;09 - 00;16;38;19 Unknown Just different targets you can use.

Again, one of the biggest advantages this target's doing what we call Target's doing. It's the thing saying, okay, I want to go 10%. How do I get there? You know, that's you're not steering with rules, but you're steering with your financial goals.

And that's also typically a new thing for for many of the brands that we onboard. 00;16;38;25 - 00;16;57;26 Unknown Mm hmm. That's a tempting feature. It's like I want to scale by 2 to 500%.

How do we get there? Yeah, but you can see also the tradeoff. Yeah, that's the thing. We also show directly in that at the trade offs, I mean, if you grow 10%, that means your margin goes down.

It means you need maybe more stock, which you have to order. 00;16;57;26 - 00;17;21;10 Unknown Right? So all of that is also there. And then you can make these trade offs because there's no free lunch.

And of course, sometimes you can actually really improve revenue and profit. That's also possible. But in many cases it's a trade off between short term, long term profit, for example, or short term profit. And that's the way these kind of trade offs which are common in e-commerce, I would say.

00;17;21;13 - 00;17;43;01 Unknown And I think when you when you make these predictions, these traders are just more visible to the to the companies and that helps them make better decisions which are carrying them better in the long term. Yeah, I love this. Transparency that you see is like if I do a B will happen. So really you can make an educated decision if we try it, you want to go, Who's your perfect customer?

00;17;43;01 - 00;18;08;16 Unknown What kind of brands do you work with? Yeah, so we work with brands starting at 25 million turnover a year. So for Germany, it's bigger customers. I mean, when I also went to the US, so there it's just a pretty small customer.

If you have a 25 million turnover, but that's more or less the case because it requires a certain amount of transaction data to work well. 00;18;08;18 - 00;18;29;23 Unknown And then yeah, we are we started in the fashion industry, but we also have customers and automotive as for iPods or pharmaceuticals, it works for everything that is sold, I would say also works offline. If you especially if you have it else, you can use it offline as well. And I think it will be more and more often used offline.

00;18;30;00 - 00;18;55;06 Unknown Of course, offline typically your elasticities are lower so you don't get as much benefit often, but it's still becoming more important because people are checking the prices. And if you are trying to set up an electronics product at the super high price offline, it just won't work. That makes perfect sense. Obviously, everyone is checking in the store if it if they can find it cheaper online.

00;18;55;06 - 00;19;12;09 Unknown So yeah, cheating doesn't really help you there because they will find out any way. You mentioned that the onboarding process, the data quality. I want to just touch on that one before we leave. So what kind of homework does immersion need to do before they can get started, before they should approach?

You know, I think really we help them with that. 00;19;12;09 - 00;19;32;21 Unknown So I mean, they can download the data requirement and look at it. And again, most people say, Hey, we have that data and I think they have it. But I think what I see on the on the pricing product is really this attribution to transactions.

What makes things difficult? Of course, you know, more or less your outbound course, but do you really know what exactly on each transaction? 00;19;32;21 - 00;19;57;24 Unknown Right. That's often difficult, but it's important.

And then we also do marketing optimization. So I mean, my performance marketing optimizations in there, again, it's it's the attribution. Like how, how do you know which campaign drove actually which traffic and ultimately which transaction. That's kind of the most important question.

I was surprised that came from pricing. Right? And I thought, okay, data quality is poor. 00;19;57;27 - 00;20;23;01 Unknown But yeah, I mean, wait, before you go to marketing and that's this was far worse actually in marketing side.

So yeah, you generally have that type of data. We're asking for, but especially the attribution maturity is very different across companies. How are you able to attribute the cost that you have with the benefit which are basically the transactions? 00;20;23;04 - 00;20;44;05 Unknown And some are very good at that and some haven't even started being in digital marketing for more than 25 years.

I'm completely with you. Digital marketing attribution when it comes to conversions is really the Holy Grail because it's so unreliable. Felix 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;44;08 - 00;21;04;27 Unknown One more reason I think why you should get into that is really that we also have to understand where the whole e-commerce segment is going, right?

I mean, there's more and more marketplace revenue coming, more and more marketplaces in itself. And I think also customers get more and more help making decisions automatically or getting support on their consumer decision. 00;21;05;00 - 00;21;29;18 Unknown So I think it's a very natural thing for retailers and brands to also use a Iot to improve their own decision making, because this is really what it is, what it is all about. It's about making better decisions for pricing, for marketing, also for ordering.

And I think it's not an optional thing. It's not like you can say, Oh, and I might do it at some point. 00;21;29;20 - 00;21;58;10 Unknown So it's really you have to do it. I think it's like the big ones are already doing it.

I mean, I think there was also a big story about Mart going big times into that topic. Amazon has been doing it for a long time as a landlord for more than ten years. I think the big successful players in e-commerce are already doing it and I think if you want to get support and be successful in that new tech world, you also need to start working on that. 00;21;58;12 - 00;22;17;02 Unknown Yeah, and your solution definitely helps with that.

And I think it just opens the doors and the opportunities for for smaller brands to use the same solutions than what you just said. The big brands are using for a long time. So it levels the field to a certain degree and everyone has has the same chances to to win the customer. 00;22;17;04 - 00;22;35;09 Unknown Where can people go and find out more about you guys?

They can approach me on LinkedIn or just find seven learning's on LinkedIn or just go to our website as well, book a demo. We also have Frank frequently webinars, but you can get in touch now these options all all good for us. I will put the links in the show notes, then you just one click away. 00;22;35;12 - 00;22;58;00 Unknown Felix Thanks so much for giving us an overview on pricing and I agree pricing is the easiest level if you do it right to get more out of your business and your solution definitely helps with that.

Thanks so much for your time today. Thank you. Also by.