Goodby Flat Files: KoRo’s Marketplace Automation #144

Another big one I would say is just data hygiene.

The rule engine is very flexible.

You can make a pear out of an apple.

But if you do not have the right source data, it's very hard to manipulate it afterward.

Let's Talk Marketplace the Marketplace Podcast with Ingrid Lommer and Vana V. Dichtech hello everyone and welcome back to a new episode of let's Talk Marketplace.

I'm Ingrid, your host, as always, and today I'd like to talk to you about a sexy word, let's call it like that, scaling.

Don't we all dream of building that cool D2C brand and blowing up across European marketplaces?

Except it's so easy.

Not always so easy in real life, I think.

So today I'd like to talk to a brand who has actually lived that and will tell us all about the ups and downs that come with it.

If you are into Halsey snacks, then you might already know them, maybe for their dried mangoes or the deep fried strawberries that my son is always swearing too.

Because we are talking about Kuro.

They grew from a hip Berlin startup into an absolute omnichannel beast.

But how the hell do you push that kind of bulk volume across Amazon and a lot of other marketplaces and internationally platforms as well without your backend completely imploding?

And to answer that, we also are looking into the engine room today.

And to pull this off, we need some technical muscle and technical expertise to explain what happens in the back end in a company like that.

And that's where Channable comes in, the software behind Cobo's Marketplace feeds.

Yeah.

So we're going to talk about the good and the bad and the ugly of multi channel growth of a D2C startup today.

So with me are Adrien Kostielli and Sebastian Thalheim from Kuro and Lucas Basa from Chernobyl.

It's great to have you all.

Welcome to the show.

Nice to be here.

Thanks.

Thank you, Helen.

Hi everyone.

It's really great to have you.

Yeah, enough from my introduction here, let's hear it from you guys.

So please quickly introduce yourself and tell our listeners who you are, what you are doing at your companies and what you would like to talk about today.

So Adrienne, why don't you start?

Yes.

Hi everyone.

So I'm Adrienne.

I started four years ago to work in the Marketplace team at Koro.

I started as an intern and now I'm a Marketplace operation manager.

So I help Koro to be live and online in a different marketplace.

Yeah, I'm really happy to be here.

Excited and yeah, can't wait to talk more about Marketplace channel and Koro.

Yeah, sounds good, Adrien.

Quite a journey you've got there at the company, I think.

Okay, Sebastian, what about you?

Yeah, thank you everyone for having me.

Also on the show, Ingrid.

I have been with Kuru now for five and a half years and actually know the time before we had Channable.

Actually I founded my own Amazon FBA business before joining Koro.

So I know the process from beginning until the end and then focus on building the marketplaces department for Koro for three years and now move to the operational innovation product team Digital product team department.

Also helping other enterprise solutions to be integrated into our IT landscape.

So I have quite some experience also what comes with this migrations, data of product data and so on.

So really excited to share some of this today with you as well.

So the real back end powerhouse here.

That's great.

Okay Lukas, that leaves you.

So tell us, who are you, who is Channable and how are you connected to those two guys?

Yeah, so I'm Lukas.

I'm the product lead marketplace at Channable.

So I work with a team of product managers and engineers to work on our marketplace connections.

All the features we have inside of Channable within the Marketplaces ecosystem.

And Quarto is one of the many successful clients who use Channable to make their life a bit easier on marketplaces using Channable to, to get their items live, to make changes in bulk and yeah, more things that they probably tell you about themselves.

So yeah, that's me.

Okay.

Right, that just leaves the big question, Sebastian.

What's the best snack of Koru?

I mean I love the freeze dried strawberries and I'm a big fan of the pistachio butter as well.

Can really recommend that.

It was top of search when I, when I looked at Amazon as well, the frozen strawberries or the dried freeze was.

Yeah, as I said, ask my son about them.

He's going to give you a whole testimonial.

But yeah, okay.

Right, so that's good.

Looking forward to talking to the three of you about the the engine room behind the success like Koro and about this great and interesting case study.

Great to have you all of an Alfred you.

Yeah, we'll dive right into it after a very short advertisement break.

Are high quality product pictures with fast production times and low production costs an issue for you?

Then you should get to know our partner, PixelModer.

The key feature here is that the AI does not generate the images, but it supports the team and models in taking the best possible shots quickly by optimizing camera position, lightning model poses, etc.

This significantly shortens production time and reduces costs up to 70% for photo production and up to 90% for video production.

This is money that can then be reinvested in the production of more visual content.

PixoMota produces over 14 million images and videos per year and operates virtually around the clock in eight locations with over 850 experts.

Its customers include over 100 brands from more than 20 countries, including three of the top five global online marketplaces for fashion.

Pixelmota pursues a unique onboarding strategy.

It begins with strategic coordination and a free trial production run with initial training on the proprietary AI with no obligations whatsoever.

The motto is let the results speak for themselves.

If you want to know more about them, listen once again to our episode 126 with Giannis Sarazi, the CEO of Pixelmotor.

You'll find the link in the show notes.

Okay then, I guess let's go.

Yeah, maybe.

I think the first question goes to Sebastian because Koro has been around for a while.

You started out as a community driven brand.

What are marketplaces to you and when did marketplaces become part of your strategy?

Obviously we started as an E commerce brand first, but there's different market marketing channels that we explored.

We were very heavy on influencer marketing, but obviously marketplaces, especially Amazon, became strategic very early on and just to also be able to get other customers that we wouldn't reach with our normal marketing channels.

So became like a strategic pillar of our strategy very early on.

Did you ever worry about cannibalizing your own?

Very profitable, as I understand it, DTC shop with the marketplace business.

Yes, of course.

I mean there's like fees that you have to cover and so on and if you outbid your own website offer, then obviously this is rather counter productive.

So how we do this is like through smart pricing.

And also.

Yeah, it's a different audience.

Right.

Like people that buy on Amazon just buy one or two products, but people that come to our website obviously have a different intent of buying and knowing the brand already.

So it's kind of like an acquisition of new, new customers as a strategy as well.

Yeah.

As we're talking or looking into the engine room today of Coral, let's look at how all of this started because you said in your introduction you have been with the company before.

You got Channable into the.

Into the frame so what did operations, especially marketplace operations look like at Koro before the this corporation?

Yes, so I think a lot of people can relate to this.

There is flat files and flat files and flat files.

So for every different marketplace they have a different set of rules how they want data to be transformed and imported.

Every country has different requirements and we spend roughly like 5 to 10 hours a week just by creating and updating listings.

There was a lot of rejections, especially from Amazon side because they are very particular of how they want the data to be structured.

So there was a lot of formatting errors and a lot of back and forth also with variant creation, which you must eventually.

But it's very tedious, let's say.

And yeah, also obviously there's orders coming in and we needed to find also a solution to find a way to import them in a good, good manner.

So it was all very, very tedious.

And also we encountered like duplicated orders imported.

So that's really a no go because then it's always manual tracking and monitoring and checking.

So yeah, there was a lot of hidden costs in this because we were like really 10 to 20 hours a week to non value adding tasks.

And this really decreased also the margin that we already had on the, on the products on the listings.

Just to understand the scale of your business at that time, can you give us some idea about how many marketplaces were you serving in?

How many countries were you active?

Can you give us a bit of a few numbers on the size of your assortment?

Because that is quite big.

So just to give people an idea of what we're talking about.

Yeah, sure.

I mean this is like 5 years ago or something.

So quite a lot has changed since then.

I mean we had like already 5, 600 products that we had to manage.

We were not in many countries.

I mean we used Amazon also as a kind of like test to go to new countries before launching the websites there.

So we were active in two or three countries and on Amazon and I think relatively quickly also on bo.com and eBay and Kaufland.

But that was a. Yeah, a lot of manual work before having Chernobyl as a tool introduced.

And did that also pull you back in your expansion plans because of all this manual work that was ahead?

Yes, of course it was a real bottleneck.

I mean we were just like, I think two or three people in the business development.

There was not even a market marketplace department yet.

So we every week had to look like what are the beneficial products to launch.

But obviously the goal was to have all products available for the customers.

But because of these bottlenecks, we were like, okay, we can only launch so many products every week and this was really holding us back in terms of how fast we can go to other markets.

Interesting.

Lukas, from your perspective as a service provider, is that a story that is common, you know, of a fast growing brand like Huo with big plans and big dreams and then being.

Yeah.

Sort of slowed down by their own manual build up that they have?

Is that a common thing?

Yeah, for sure.

I experienced this myself as well formerly, being on the HC side and setting up accounts with flat files.

Indeed.

And it's, it's not just unscalable, it's also very, very frustrating because you need to figure out things as you go.

There's not.

So you have the help centers you can rely on, but it's just something that's always going to be with some trial and error and it sets your ambitious.

So most clients have big ambitions.

They want to do three to five marketplaces and then they sort of get stuck in the first one because of these inefficiencies.

So it's definitely very common.

And I think Coro has the benefit of being sort of a retail product.

So the packaging is good and optimized already for selling in multiple locales.

But if you're just like a very local seller and you also have to do that sort of work like translations, it can be even more tedious.

Yes, for sure.

So what is the most underestimated complexity that D2C brands are not looking at when they are going for internationalization, from

my point of view, and I want to know if Gordo can confirm or deny that.

But it's also underestimating the challenge with regulations.

For example, if you're selling food or health supplements, you're also encountering stuff like SDS sheets, hazmat things you need to fill in.

All working via sheets as well.

And the problem, especially with Amazon, is that if you upload it the first time and you do it incorrectly or insufficiently, it becomes harder to change it afterwards in some cases.

Right.

Because they just sort of.

Yeah, it will work against you for some reason.

And we see a lot of users that have big ambitions and then they become a general client and they want to do these three to five marketplaces, they still get stuck because of regulations or things they just hadn't foreseen.

But maybe Goro has some experience on that as well.

Would you agree that regulations mostly or what's your most underestimated complexity at that point?

Yeah, I mean, maybe Adrian, you can take the question.

Yeah,

I would agree on this.

Certification, authorization to sell and some documentation that you have to provide Amazon, that is quite a pain sometimes.

And I guess that's also the category of our product which is food is quite challenging sometimes.

We also have not issues but challenge with best before date.

So when you do fba, you have to send to have the proper data and valid data that Amazon can use to basically have a better customer experience.

But I think in the beginning the main issue was really this flat file, the requirements of each marketplace that change over time and that are really, sometimes they are really specific to the marketplace.

But I guess everyone working in marketplace knows that it's part of the industry that requirements change.

Yes,

if I understand it correctly then Kuru is a company who has mostly built their backup themselves, is that correct?

So was that a change for you or also a cultural leap for you to say, okay, I think we would be doing better with an external partner in this area?

Yes, I totally think so.

I mean we want to build everything ourselves if we can.

But obviously if there's a solution out there that already established and has the solution at hand and a partnership is presented and obviously we take this easy, this decision and I think also with the resources at hand at the moment we were like two or three people.

We wanted to scale big time.

We couldn't just hire five engineers back then.

I mean, we had a couple of engineers working on the shop.

We didn't have the capacity to build a tool like this in house.

So this was the reason why we reached out and had a look and saw that the product Cheneba was offering was already quite good.

So this was a very easy decision.

And also it was very quick to implement.

There was a plugin, we could test it and then just go from there.

So it was really easy.

And then also always the question of scalability.

If we really want to go to new marketplaces.

Every time we would have to ask the developers to create a new custom implementation for a new marketplace.

And they couldn't be on top of all the changes because they are not experts.

Right.

So I think for some problems it's really good to work together with a solution partner who's really specialized in what they can offer.

So that was the situation then and the way that you reached the decision to go for an external partner on this matter.

So let's look at how it is now.

So we have different marketplaces.

In the beginning you told us that you were active on Amazon, ebay and Kaufland, I gather it's more by now.

So what is the situation now on which marketplaces are you active and how do you adapt your product content to each of these specific platforms?

Yes, so we are on a bit more marketplace.

Indeed.

I think we have more country for Amazon.

We're in Italy, France, Germany, Belgium, recently also and other marketplaces than Amazon.

We are still on ebay, Kaufland in France.

We are planning to launch this year Decathlon which is also integrated with channel.

And this week we launched also a TikTok.

Oh, interesting.

Which could be quite promising.

Promising.

I hope so.

Let's see.

There was quite some surprise.

For example last year we went on Galaxies in Switzerland and it went pretty well.

It went quite strong.

So we were not expecting such a, a success.

So yeah, we are still planning to go on more marketplace.

And yeah, then the question on how do we adapt our content for each marketplace?

I believe we have like Amazon focus, so we optimize for Amazon which is different than our webshop and then we go for like minimum requirements.

So we want our product online on this specific marketplace and then we really want to improve the listings by enriching the listings by adapting more the content that we already have typically for Amazon and then to adjust it so with some rules or to do some checks with Channable rules to understand where are we missing some data or is there some duplicate image or things like that.

So yeah, it's a constant work, like lots of testing also to test what kind of titles would work for this, like shorter or longer or.

Yeah, it's a constant work and since it's changing all the time, these requirements of each marketplace we have to adapt also.

So yeah, we are evolving and we use the tool channel that really help us to also grasp where we are lacking information or specific content.

Maybe I can jump in here also really quickly.

I think you mentioned this rule engine I don't know if everyone is familiar with this is just basically you can take the data that you have and then transform it with a set of rules like take a description and put the title together.

So you don't have the content team that needs to create something totally new for a marketplace, but you can basically work on your own to make it ready for the marketplaces.

Right.

So that's I think very strong for Channable that we are able to not rely on external teams, but we can reuse the content that we have.

Obviously we have some specific content for high impact marketplaces but for others if there's a field of oh we need a taste here and we don't know the taste.

We don't have this information.

Then we can easily inject it without needing to rework the whole data model or something like that.

So it's really powerful in this sense.

And then I think also because Channel has like a kind of a quality step before data is pushed to the marketplaces to kind of like check is all the required fields there.

So even if I push, they have like kind of like dry run.

Right.

You can see would this flat file be accepted or not?

So I think that's a great part here and that allows us to also expand quickly to these new marketplaces.

Okay, Lucas, I actually wanted to ask you what Channel is doing for.

For the retailers, but I think Sebastian took your part.

Yeah,

that's not fine.

No, that's.

I mean that's the best that you can do actually.

But Lukas, can you add anything to that?

So what are you helping with and what happens with the data when it's with the retailer, this data when it comes into Channel?

Yeah, I think Sebastian did a good job, honestly.

Basically, we always have this sort of the marketing terms is we enrich and allow you to enhance your product data.

But a good example, a very practical example is you are a German seller.

You have all your content in Germany, color blue is blau, and you're trying to expand to a Dutch marketplace and they won't accept blau.

And with Channable, you can set up one rule, translate everything in automatically so every blau becomes.

It's actually pretty much the same in Dutch, but you get what I'm saying.

But the colors are converted.

You can do the same for materials, even sizing structures.

And by doing that inside of Channable, you don't have to clutter your ERP system with a color field for every locale that you're selling into.

Right.

So you can do it all with these rules.

And these rules are copyable.

So if you then go to another marketplace, you can just copy those rules over, have everything set up for.

For the Dutch marketplace.

So I think that's the rule engine is.

Is the most.

Yeah, the most unique thing within Channable in terms of marketplace, sort of making that easier and it takes some time to fully grasp because you can do some crazy stuff with it.

Almost like half coding.

But yeah, this is the gist of it, I would say.

Interesting.

I have to look a bit at the time, but I would like to dive into one of those crazy things maybe in a bit more detail.

Let's look at maybe the price, the area of pricing.

Because I figure this is an interesting and important part for Koro, just because you are in the food sector and coming from Germany where this is.

Yeah, the price is an interesting feature here and a very important one as well.

And then we have the part of, you know, fluctuating raw materials and stuff like that.

So I guess dynamic pricing is a topic for you.

So how are you handling this, this whole pricing situation across various marketplaces and across various countries?

Yes, so we indeed use dynamic pricing through again this rules feature, which is so we know the raw cost and we also cross check against our shop, as we mentioned before, to cover the marketplace fees and the change in price.

So it goes through rules and we have different country price.

But the reality is that sometimes we are not the cheapest one on the marketplace.

I mean we are the cheapest seller, but we are not as cheap as on our website.

So sometimes we do lose a buy box in Amazon, for example.

So yeah, we do sometimes do manual adjustments on the pricing, but most of it goes through pricing rules that we've set up in Channable directly.

So when our cost change, the price also adjust accordingly.

Can I ask a question on that?

So you're quite well known in Germany.

Is the pricing strategy different in Germany from another country where you're still emerging a bit more?

Prices are country specific, of course.

And each team obviously has their own strategy how to enter a market.

And yeah, this is also reflected in the strategies, of course.

I mean we just, we cannot copy anymore.

We are like on the stage of localization, we don't translate.

So yes, of course,

as price is a bit of a sensitive thing.

I won't ask you for specific examples here on the coral side, but maybe Lukas, you can help us out because Adrian mentioned that they are using this rule based model that you are offering to go for different prices and create a dynamic pricing for different marketplaces and countries.

Can you give us some examples on what this is capable of?

Yeah, so I think there's two options here.

We have a Repricer which does it based on the buy books price and sort of helps you get the buy books automatically.

But with the rule engine you can also manipulate pricings, for example, by saying if my stock drops to a certain threshold, I want to increase my pricing a bit to prevent selling out.

That's an example.

Or you could take a specific category and increase prices there for a period of time because I don't know, stock in general is low.

So yeah, there's multiple ways you can manipulate Pricing, and I would say with rules you can do it based on.

I think a cool example to mention is we had a different client and they linked to the Weather API, which does the weather, and they imported that and they had a rule that said if the sun is out and it's above 23 degrees out, I want to lower the price of my sunglasses and push them a bit more.

So that's sort of the cool stuff you can do.

Very cool.

Yeah.

Okay.

So, yeah, we have been talking about things that are possible and are working well, but as I promised at the beginning, I'd like to talk about the ups and downs as well.

So if you are looking back at maybe the last two years for coral scaling on marketplaces, what was the biggest structural challenge or a miscalculation maybe on your part and how did you go to fix it?

So I think the biggest challenge for us was, I mean, for me at least, was to make sure that we have stable connection, like stability across all our systems and also to be able to import orders and update listings fast.

And I believe that's what Channable brought, especially on the order import side.

We were able to have, once an order is made, we have like a half an hour Amazon checking for the payments.

And then with the previous solution we had, the import was taking too long, like more than an hour.

Now we are right on point.

When the order comes in and is validated by Amazon, it starts flowing immediately into our system.

So that really helped us to get the prime badge, for example, and make sure that we deliver on time.

And I think we got much, much better at this.

We have our older digital product team that also work on this connection between the systems, erp, warehousing marketplace, obviously.

So, yeah, mainly stability and the speed to which we can integrate and go to new marketplace as well.

So stability and speed.

Lukas, would you add anything as a general structural challenge that you often see in your customers or maybe even see when new customers come to you and start out?

So what is the main problem that they bring?

Yeah, so I very much agree with the stability factor and reliability.

It's a big topic here as well, because ultimately that translates the quickest into the downfall of an account where you're unable to do orders in time.

Another big one I would say is just data hygiene.

So the rule engine is very flexible.

You can appear out of an apple, but if you do not have the right source data, it's very hard to manipulate it afterwards.

So that means having a set amount of attributes for all your items and also the consistency that for new items, you also fill in these essential fields because that will allow you to set up these automations that will transform everything for the channel without you having to touch a button.

But if you then have an item that sort of omits all the important stuff like a description, etc.

That can sort of crash down a bit as well.

So I would say for sure, data hygiene and then reliability is definitely up there as well.

Yeah.

So what would you say?

How much time should merchants invest in this first setup phase?

Because when I hear you correctly, then it pays off if you put a lot of brain cells into that process and really think about how your products are going to be structured, because this is something that will save you a lot, a lot, a lot of time later.

So, but because we always talk about quick onboarding, but is there a too quick?

I would say so, yeah, I would say there's definitely a too quick.

And we try to prevent issues with stuff like a dry run where you can do a test setup without actually sending the stuff to the marketplace.

But if you cut corners, eventually you'll have to fix the stuff that you've made concessions for.

So I would say you have the onboarding phase.

And indeed, you know, as a channel, we want to make that as short as possible with having all these, these, these cool features.

But there's, I would say, a period that goes before even the onboarding process, where it's sort of the preparation phase, where you think about, okay, what channels do I want to actually do, what products do I want to sell on these channels, and how do I get the right data and the right regulations fixed so I can actually start selling.

And too many times we encounter sellers who've done all their setups and then they find out they are unable to sell a specific category or a specific brand or they don't have permissions.

So that's doing it backwards.

Yeah, absolutely.

Okay, to come to a close, let's have a look at the future then.

So what is Kuru aiming for at the moment when it comes to their marketplace business?

What are you preparing for?

What are you planning?

And yeah, maybe also let's touch a bit on AI and how that changes maybe this future outlook that you have, Sebastian?

Maybe.

Yeah.

So obviously we want to launch in new marketplaces, be everywhere where the customer needs and wants us to be, have like a broader omnichannel presence overall.

Obviously, marketplaces is a big part of this.

And yeah, I think when it comes to AI, also, I think what is really great now that we can also work with like reviews in a better way.

I think that's one big pillar that we want to integrate as well because it's a lot of text.

Right.

But now we are able to also understand the sentiment of it and integrate it in our product development to actually react to user feedback and not just like do our best guesses, let's say like this.

So yeah, and then obviously become better in operational efficiency, be faster, be direct and yeah, put, put out the content that actually drives the conversion rate because the informations that the customers need to make the decision for them is on point.

Absolutely.

Lukas, anything to add on the matter of how AI is going to change how you are tackling marketplace business in the future?

Yeah, I would say it will lower the time needed for you to go live on a new channel.

The Flat Files era is luckily far gone now, but it will become easier to set up a new channel and to experiment and validate and there will be more time for actually performance and making sure you sell more items.

And I think it's super to facilitate that as a tool and to grow along the market basically.

And what I'm very curious for as a, as a sort of Amazon seller myself, previous Amazon seller myself, is that retail media will also be impacted by this whole new agentic wave.

So how will that change on bigger channels, but also eventually smaller channels?

It's very interesting and something you have to keep track on as a marketplace seller.

I would say.

Now that's a whole different episode I think.

But yes, definitely, let's keep an eye on that.

Okay.

Maybe to close.

I always like to close with something of a quick win for our listeners.

So let's pretend it's next Monday.

And if merchants listening to us today want to see a tangible improvement in their marketplace operations or margins by the end of the week, what is one lower hanging fruit that you could give a strategic tip on?

Coro first?

Maybe.

Yes.

I like the idea of creating bundles like when you have or virtual bundles when it's available to combine two great selling products or to help customers discover new products through good selling products and maybe more new and innovative ones.

So we'll go try, go, go try the virtual bundles for example.

Okay, maybe from my side go and use, I mean Channable if you have it or any other tool and check where is the description too short or where are missing bullet points, where is actually data missing and then start a project on filling this in because this is the biggest driver I think good data if it's an agent buying in the future or a potential customer.

This is, I think, the cash cow.

So, yeah, that's something you can tackle starting next week.

Yeah, absolutely.

Lukas, one from you.

So I would say maybe a bit of an open door, but if you've set up a channel and you've went through all the hardship to set that channel up, don't stop there and add a second one.

So if you're selling on Amazon at Kaufland or Otto, and every locale has multiple marketplaces, and the sort of the return of investment is much higher because you can copy a lot over.

You can even go to Marketplace, download your catalog, import it into Channable or another tool, and push it to a new marketplace.

So I think your investments will have a better return of investment or return on them.

Basically, that's what this channel here is all about.

Thank you for that, Lukas, and thank you, Sebastian and Adrienne, for joining me today.

That was interesting.

It was great to have a look at the story of Koro on Marketplaces and especially look in the engine room for a bit and look how this really works out behind the scenes.

So great that you joined me today.

Thank you.

Thanks.

Thank you very much.

Yeah.

And if you like listening to us today, you could also always make us very happy by following us and rating our podcast on Spotify, Apple Podcasts, YouTube, you know how it goes.

Just, you know, click that, like button and I'd be happy.

And if you want to tune in again next week, where we have another great case study coming up, then it will be about customer service and how to do that on multiple marketplaces.

A whole different beast.

Yeah, we'll talk about that next week for today.

Thank you and bye.

Bye.

You listen to let's Talk Marketplace, the Marketplace podcast with Ingrid Lommer and Valerie Dichtel.