Ecommerce Data Demystified: 5 Insights for Success | #218 Feifan Wang

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 @clauslauter 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.

Claus Lauter: Hello and welcome to another episode of the e commerce coffee break podcast. Today, we want to focus back on the topic of data. Now data for a lot of merchants is part of their daily business and it's all over the place, but we want to find out five data challenges that plaguing e commerce businesses the most and how you can overcome them.

With me on the show, I have Fei Wang, he is the CEO of Source Medium, and he's a seasoned entrepreneur with a rich background in sales, software, engineering, and product analysis. He was born in China, immigrated to the US at 14, and his early experience of cultural adoption laid the groundwork for his entrepreneurial spirit.

He worked for Google, followed by a VP of product and analytics role at Residence Home. There he navigated the company's rapid scale up and complex data challenges. Gaining unique insights into the struggles that brands face with data management and these experience finally inspired him the creation of source medium, a leader in e commerce data software.

So let's dive into the world of data and welcome Faye to the show. Hey Faye, how are you today? 

Feifan Wang: I'm good. I'm good. How are you Klaus? I'm 

Claus Lauter: very well data. Part of every business. very complex. Specifically if you grow your business, might be easy if you're a smaller merchant, a smaller enterprise, but once things become bigger and bigger, things become much more complicated.

You help us that give me a bit of an idea. Why. Data can be such a burden, can be so complex. 

Feifan Wang: I think the most important thing to understand is that it evolves based on what stage of business that you're in.

 If you're in the stage where it's really still looking for product market fit. Making sure that the product is actually liked and enjoyed by a large enough audience and things like that. The challenge there really is around just having something right. GA is usually sufficient. You just need directional insights to know if you're doing something good or not right.

But once you get to the stage where a lot of these things are figured out, at least 80% of them are figured out and you're now really thinking about the medium to long term in terms of scaling, right? Whether that's scaling your top line or optimizing for profitability, then the data that you're working with becomes a lot more important and I think an analogy that I like to give here is right.

If you're driving at 20 miles an hour, Your numbers off 20% either direction isn't going to be that big of a deal, if even noticeable at all, right? But if you're driving down the highway, that becomes almost life and death sometimes, right? So that's how I would think about it. 

Claus Lauter: Now, with a growing businesses, as I said, you get great data from all kinds of sources.

Now, some data is hosted in your own system. Some data is hosted on external systems, and it becomes really, really difficult to find out how to compare and how to read out the data. Now at SourceMedium you do it slightly different. Tell me a little bit on what your approach is. 

Feifan Wang: We're ultimately a data company that Is about the data and not as much about the user experience, right?

So I think most solutions that you see out there will have a lot of focus in terms of their marketing on the visualizations that they have or integration coverage or things that are perhaps more UX oriented. In terms of user ease of use, but what we really focus on is to be the provider of the highest quality data in the commerce industry, right?

So in terms of how we think about data quality, there's a few things. 1 is around accuracy, right? Is there out of the box accuracy in the sense that do your numbers match with the what platforms actually report on and you? But 9 out of 10 times, once you have some complexity in the business, the numbers on the platforms are actually also incorrect for various reasons.

So does the data provider have those nuances built into the product? So that's the accuracy piece. And then the next piece is around breadth of the data. So that's just essentially the integration coverage That it has, right? Does it sufficiently cover all of the different various platforms that you're going to be using as you scale the business?

And then lastly, the depth of that data, right? How rich is the data that's actually coming out? Right? Because the richness of that data is very important in terms of how actionable that actually becomes. As well, so that's kind of like the main thing that we focus on and then the operators that we cater to our folks that are a little bit more sophisticated in terms of what they're looking for in the quality of the data and the type of decisions that they make that they think has to be supported by analytical insights and then we have the A couple of different types of consumption methods.

 We have brands that are consuming our data directly in our hosted visualizations, or we have more sophisticated enterprise customers that are consuming our data directly within their data warehouse. 

Claus Lauter: Let's dive a little bit deeper into quality of the data, because I think that's a bit of a gray area for a lot of merchants out there, because they assume that whatever data comes in should be right, but probably it's not.

So what does quality of data actually stands for? 

Feifan Wang: That really is a little bit different depending on the phase of the business that you're in and what. Metrics your what KPIs you're orienting around so besides the high level understanding of accuracy, breath and depth, There is also, for example, just to out of the box, being able to have reliable net sales. Based KPIs, right? Let's say if you base your CAC on that sales or your LTV on that sales that has a lot of implications there. For example, you have to have the correct gross calculation and correct discounts and refunds calculation.

And that also differs across borders, right? So the Shopify data that comes out of Europe, for example, has certain implications that's different from the states. , as the brands grow and they get to a point where they want to, for example, optimize around profits, then that can open a whole can of worm, right?

Because the expenses data and the cost data. Are very fragmented and there's a lot of different opinions as far as how certain things should be calculated. Where are you pulling those numbers? Accuracy, the way that I think about it is you can never have 100% accuracy. It's almost physically impossible.

Because of just the levels of layers and complexities that exist, but also you also get to a point where you enter into the world of opinions. So that's where a data culture becomes very important. Having a standardized taxonomy within the organization, having a source of truth that everybody in the company is orienting towards is extremely important.

For example you could. Calculate your ROAS based on your growth sales or your net sales. Well, which one do you do, The marketer may prefer to look at it based on growth and the finance person may prefer to look at it based on that. And the CEO may prefer to look at both, but really they're tracking one of those, right?

So you also end up getting to that where it's just about the human beings and the governance that's coming with that as well. 

Claus Lauter: A very good point that you mentioned there. I've been sitting in a lot of shareholder meetings looking at the same data and everyone was basically reading something different out of it.

Now, obviously you use the data to make your decisions based on that. And I think visualization is a point there. And that's something you do and you help with. So how do you visualize all this data for the different players in the game? 

Feifan Wang: Great question. Yeah, so our visualization design philosophy. From the get go has always been orienting towards the more sophisticated operators

 when you kind of look at our hosted visualization templates. It can be overwhelming for some, right? So we've seen other solutions where, for example, it's really about the simplification of that, right? Go into scoreboards, very simple graphs and lines and stuff like that. But our product, when you open it, it almost looks like a Bloomberg terminal to some.

 So it takes a lot of intentional learning and onboarding with us to really understand how to look at it. Right? But of course, that granularity and richness of data can only be leveraged if the human be on the other side is already committed to working that way. Right? So ultimately, we're very.

Clear about not everybody is the right fit for our solution. And then I think for folks that are consuming our data directly to build their own visualizations we're more than happy to provide ton of advice there in terms of. How can they connect the dots between our data and the charts that they wish to build and of course, anything that we've already created ourselves would be able to help them stand that up as well within their own BI of choice.

Claus Lauter: lot of companies or probably every company has a different data stack where they get our data sources, where they get the data from how do you integrate with all these? Different data sources and give me a bit of an example on what's the starting point and best case 

Feifan Wang: examples. Great question.

Yeah. So we are a vertically oriented data infrastructure solution. So then the question might be, what is a horizontal data infrastructure solution? Right? So I think the way you can think about is like the super metrics funnel 5 trend of the world, right? They're there to serve Basically, a lot of different industries.

So they have a particular boundary where they won't go further. With the solutions that I mentioned, their only kind of mo is to move raw data from A to B. but that doesn't really, that's where your problem begins. Because there's so many other things that kind of comes from that.

And, but for us. Not only do we take care of data movement and extraction from your data silos, we also have very advanced data transformation technologies. In terms of how we actually activate that the reason that this works for commerce is because there has been an aggregation of.

Customer base and user base into finite set of platforms, right? So if you think about why this is, this will be way harder for SAS is every SAS company have their own schema, have their own database model, right? I have their own intricate sort of ways of operating, but, most merchants are on especially brands that have started in the last few years is

finite number of platforms, right? You have your e commerce transaction and customer data in platforms like Shopify, Magento of the world, right? You have your customer communication data as support data in platforms like gorgeous Zen desk, Klaviyo, attentive, right? You have GA, you have your marketing data from Meta and Google and Tik TOK, et cetera.

That actually makes it possible for us to really orient towards a set of platforms that covers the 80 20, as far as what the brands are using, and then essentially creating a standardized taxonomy. 1 of the things that's really powerful with our offering is that it's omni channel by design.

So what that means is if you just understand, for example, how net revenue is calculated for Shopify, you'd be able to understand that for any Amazon business. For anybody on stripe on charge B, but really on any other e commerce platform that we would integrate with down the road. So that's really what people love about our offering is that you just need to really learn the taxonomy 1 time.

 That becomes impossible with SaaS because every SaaS companies offering is just could be so, so wildly different. 

Claus Lauter: Yeah, I can totally follow that train of thought. Obviously was a technical stack Shopify, Klaviyo, Gorgias, whatever you made it a lot of merchants, a lot of listeners will find themselves in that kind of ecosphere, and then it's relatively easy to do the next step.

No. What's the kind of homework or when do as a merchant, do I need to think about getting up to the next level and maybe getting contact with sourcemedium. com to get my data 

Feifan Wang: consolidated? We have this concept internally of the data maturity curve. For example, in my last company, I got to participate in the scaling of the brand as we went from seven to nine figures.

And now beyond, the biggest online mattress retailer in North America now. Started right under 10 million in sales. I remember, but each stage of that journey required a certain amount of self awareness in terms of what it is that you actually need. How do you think about that?

 In terms of the maturity curve, it's, made up of how large your business is, but also how complex your business is you could be, for example decent sized business, but you're only advertising on Google meta Shopify. One could argue it's not as complex, although that Is very rare, but you could be, for example, a 10, 000, 000 dollar brand that's selling on retail Amazon Shopify.

 You may have more than 1 brand. where you may be selling in both the US and the UK. All of these things are going to very rapidly. Increase your complexity level, regardless of your actual revenue size. That's the most important, but for every business, they will get to a point where their spreadsheet set up is not really doing it anymore.

 Whether that is trust of rely, a lack of reliability, or it's just not loading anymore. Or maybe the out of the box Shopify app that you installed is, it doesn't have the richness of the data that you require to make the type of decisions that you need to be able to make. 

Or you realize. That the numbers that you're seeing is, let's say, 20% off, then what's actual, right? But now you actually care. Whereas before, maybe you were fine with that, right? Because it was smaller. So once you get to that threshold, then it's time to really think about this concept of a data strategy, In terms of what are you going to be adopting that is going to not only support you today, but also the next 234 X of growth that you're going to experience, right? Because the last thing a lot of people want to do is probably to get to another threshold in six months and you have to do a whole other round of vendor evaluation and things like that, right?

And that can really, really slow down our business. If you do that. 

Claus Lauter: Yeah, it makes total sense. Obviously some businesses are growing slower and more predictable and some business and hopefully you're among that as a merchant have this kind of ice hockey stick. Grows, which for a lot of them comes with a ton of problems and challenges.

And then it's better to have the right setup already done. Now you spoke a little bit earlier. You mentioned is the onboarding process where you help with data quality and all of that, how does that work? 

Feifan Wang: Yeah, great question. So for us, it's pretty straightforward. You sort of just all off into your various platforms and then we ingest the data that's.

Pretty much the same with every data provider out there. with our onboarding because of our focus on data quality, we actually do a preliminary data quality check. Making sure that our numbers are matching, right? For example, what's coming out of the shop by sales report, And if there are differences, we actually spend the time to, understand that ahead of our actual onboarding call, That's 1 layer, just like numbers matching. So that's very easy to understand, but. Then there is a lot of other data hygiene things that we help identify proactively.

For example, you have a lot of non nons or direct nons in your UTM values, or let's say you're not tagging your UTMs correctly, or you're not. For example, using something like UTM campaign that can give you some richer information there, right? Where you're using inconsistent naming conventions or discount codes are being used in inconsistent ways, right?

Or let's say for a 0 orders, the CS department is doing different types of things that actually makes cleaning that data more tricky. So all of those things we would be identifying proactively, but also what are some of the things that we know that is a generally accepted best practice that the brand is not doing.

So an example that I can give there, for example, is not doing post purchase surveys. Getting information on how did you hear about us? we integrate with our partners in the ecosystem there as well. But I think ultimately the reason a lot of people haven't done some of those things, even though they may already know the importance of it is they haven't had an actionable way to really look at it.

the aha moment that we typically get is Okay, well, I can actually use this insight now I can actually make types of decisions based on UTMs that I didn't even think was possible. And then that opens up a lot of other thoughts there right I think just being able to look at your last click attribution relative to the zero party attribution, aka how the customer heard about you.

Becomes very insightful, right? Because you typically see, for example, a lot of Google taking a lot of last click credits, But then you might say, you might actually see that. Well, they actually heard about you on YouTube, or they actually heard about you on classes podcast or what have you. So all of a sudden, it's like you're going into high def, right?

Whereas previously you're just Suffering under the obscurity of Google CPC source medium, and having a lot of opinion based debates around. Are we overspending on Google, or how do we think about this? You really want to just be moving your discussions from the land of opinions.

To a land of consistently analyzed at consistent analysis based on a trusted source of 

Claus Lauter: data. Awesome example that you gave there was the last click attribution. All the providers out there always want to have the last click attribution because they want to claim the business or the click for themselves.

And that screws up the data. And it shows you also that having a good structured system and an external partner who knows on how to look into data will help your business massively. Now, who's your perfect customer? Yeah, we 

Feifan Wang: typically work with brands that do sales between 20, 000, 000 to up to a billion.

But with that being said, we also work with brands of all sizes, because we have a very selective few agency partners that are able to essentially use our data as their infrastructure. And service that to their own clientele. but we actually also work with SAS companies who take in our data to power their own reporting or power their own feature sets.

And of course, then we indirectly serve those customers as well. would say, in terms of brands that are coming to us directly and we're working with directly, there really has to be a commitment to onboarding with us and stepping into our methods that also requires work on the brands and in terms of, let's say, cleaning up things like UTMs, right?

Cleaning up internal practices to make sure that the data coming in is going to be standardized and clean, right? Actually learning about some of these neural pathways, going from, let's say, having an analytical question to getting to a trustworthy, actionable piece of insight.

But then I would say, in that case, it's typically brands that have some level of complexity baked in apparel is a great example of that, right? They have a ton of exchange data, reshipments, influencer samples, right? Tons and tons of orders that you really don't want to be calculating.

In your metrics, Or let's say they have Amazon and Shopify, right? Or let's say they're leveraging a retail partner like leap, right? That's sitting on top of their Amazon store or tap car. And you need to be able to look at the performance of those channels separately, right? So all of those things would be just like out of the box.

For us, that's not for every merchant, Ultimately, the long tail majority of merchants , are probably just on Shopify, probably just using Google ads and meta and probably just having GA, so in that case, we could obviously absolutely still work for those brands, but it's really for brands that already have some sense of the complexity that's either incoming or that's already in their business.

Claus Lauter: I like the focus that in your onboarding, you basically completely dissect what they have and not only provide a technical solution, but also the expertise on how to make it better. And I'm sure you must have seen a huge mess with one or other clients onboarding you. I can just imagine that some people just do not have control over their data.

Give me the overview about what's the pricing structure, just a rough 

Feifan Wang: idea. Yeah, so it depends on the use case so we have more of a wholesale model for agencies and SAS companies. That's more so as a potential replacement for products like Fivetran that they may be using. But direct to brand, we look at their training, 12 months sales, and we have different sort of levels there, depending on how much sales that you're actually doing.

That's actually flowing through our system. for example, the decision to say, whether or not we want to integrate Amazon becomes relatively easy because how much sales is coming through that. And how much increased cost and that was be bundled into that as a 

Claus Lauter: result.

Okay. Before we come to the end of our coffee break today, is there one final thought that you want to leave our listeners with? 

Feifan Wang: Thank you for allowing me the time and the airspace to talk about source medium. The final thought really is to reiterate this relationship between understanding where you are in terms of your own maturity as a brand.

And having establishing opinion between the complexity of your business and the needs that you actually have when it comes to data, and then being very intentional about selecting the right solution there. And of course, I think there's always the build versus by discussion as well. So it's really important to also understand if one were to build internally.

 What are the implications and scaling challenges that they will inevitably come into play as the businesses grow. There's hours and hours of conversations we can have about that. But I would say happy to chat with anybody that's in the listener base of the podcast here.

 Feel free to just email me directly at feysourcemedium. com. That's F E I at sourcemedium. com. And we'd be happy to dig deeper into some of those topics. 

Claus Lauter: Excellent. I will put the links in the show notes as always. Then you just one click away. Thanks so much for giving us a very in depth. Insight into data, why it's so important and what you can do to make it better.

And I hope a lot of listeners will get in touch with you. Thanks so much for your time 

Feifan Wang: today. Take care.