September 6, 2022
13 min read
Subscribing to different online and mobile services has become commonplace, but what about subscribing to physical goods? How does it differ from subscribing to online content and how difficult is it to start such a business?
In the 17th episode of the SubHub podcast we talk to Andrey Rebrov, the CTO & Co-founder of Scentbird. He shares his experience in starting a perfume selling company that works by subscription, behind the scenes of the financial processing, and major differences between subscriptions to physical and digital goods.
Listen to the episode in Russian or read the most interesting parts below in English.
Nikita Maidanov: This episode is going to be unusual because we used to talk to guests from the mobile market, while Andrey has a completely different, but nevertheless, very interesting story. Please, tell us about Scentbird and how it all started.
Andrey Rebrov: Scentbird is an American perfume and cologne subscription service. It all started for me quite interestingly, unusually, because I’m actually a developer. I worked for large enterprises for a very long time. But once I joined a small consulting company, where, at some point, one of the clients asked me to conduct an internal hackathon. I liked such things and so, by accident, I ended up at the AngelHack hackathon in Moscow. I met a couple of interesting guys there, we came up with some kind of an app, and took seventh place – it was fun. And then two months later, one of the hackathon organizers, Sergey Gusev, approached me and said that he and his business partner Masha Nurislamova were looking for a technical director in a startup that would help people buy perfume online with less risk. And that’s how it all started.
It was August 2013, but we were far away from the subscription model back then. The first year we used the Try Before You Buy model. We knew that a similar approach was used by Warby Parker, an American company that produced spectacle frames and glasses. They didn’t build their business through a usual for their market model, instead they became a vertically integrated company that came up with a fairly unique model: they sent their clients up to five eyeglass frames that the person could try on. However, they didn’t have to pay the full amount at once, only some kind of small deposit. The customer could take what they liked and send back the rest. So, inspired by this approach, we began to make the first version of Scentbird, which worked exactly the same. That is, we sent up to three bottles of perfume along with the samplers that we poured in Sephora. One Sephora place even banned us from getting in, as we would come there every day and collect up to 10-15 samples.
Nikita Maidanov: Did you personally come and poured samples at Sephora?
Andrey Rebrov: Yes, Sephora is a safe place to come if you need perfume samples. They’d pour you a small bottle which is enough to use for two or three times.
But apart from the samples, we also had a perfume recommender, which would select perfumes for the client based on their preferences and tastes. And the development of this recommender was actually another reason why I was interested in joining this startup. Machine learning was getting popular at that time, so I decided to hop in with creating the algorithm.
But after a year of trying to make this business work – it didn’t. It didn’t work out in Russia, due to mentality and cultural aspects – sometimes the bottles were simply stolen, which was not the case in the US, though. And then, after two and a half months of being in the accelerator, every mentor would tell us how stupid our model was and that they, and their friends, would never invest in us. So we had to change something.
So we started having deeper interviews with our clients and noticed a recurring thing: people have five or six half-empty bottles of perfume at home. They want to try new scents, as every year there appear about a thousand new titles, but they feel guilty of having the previous bottles left unused, especially considering that one bottle usually costs $100-$300.
That’s how we came up with the idea of repackaging perfumes and sending people only a monthly supply. Sounds simple, but there was always a risk of infringing copyright or a trademark. Even our mentor confirmed that should we become a little more visible on the market, the brands would start writing to us asking to shut down.
Vitaly Davydov: You mean you’d just go to a store or a distributor, buy a large batch of perfume there, open all the bottles and pour into your “White Label“ bottles?
Andrey Rebrov: Yes, that’s exactly what happened. We bought from wholesalers. At first, when there were very few subscribers, we couldn’t even open the lids properly, had to use the pliers. But usually we re-sprayed the perfume from a big bottle to a smaller one, because if the bottle is opened, all the spirits and essential oils start disappearing. Depending on the formula of the perfume, they fade at different speeds. But as we expected, a year later we began to receive the Cease and Desist letters from the brands, stating that we must either remove their perfume from our site or we’d be sued for trademark infringement. Some scents we had to remove, but some we managed to keep by coming to mutual agreements. So we were caught somewhere in the middle. But some time soon, brands began approaching us on their own and we started signing official contracts. So by now, the majority of fragrances presented on our site are provided by the official brands.
They usually bring a tank with the fragrance to our storage in New Jersey where we pour the bottles and distribute them further, everything is automated there. But when we need to cover a really huge amount of bottles (hundred thousands), we come to our partners, who are professionally engaged in bottling perfume.
Vitaly Davydov: Awesome!
Vitaly Davydov: Could you tell us more about the accelerator you were in?
Andrey Rebrov: In 2014 we passed the ERA accelerator – Entrepreneurs Roundtable Accelerator in New York, we were the seventh batch. And we also got to YC in the summer of 2015. We were far from being the first Russian company there. But it didn’t mean we knew how everything was going to be like. There were no groups in WhatsApp yet, there were no Russian-speaking YC communities. In fact, before we got there, I didn’t know much about making a startup. When we started at YC, I didn’t know a huge number of things, up to how to register a domain and how to link it to the site. Because I had been engaged in enterprise development mostly.
Vitaly Davydov: Was it difficult to decide to switch to a startup that works with fragrances from enterprise development?
Andrey Rebrov: Yes. The lack of a certain financial future was tough. But I wanted to, it was interesting. It was interesting to try something like that. It was interesting to create something. That is, again, one of the reasons, because I went into development, I like to create things from scratch. It was interesting to create a startup. It was interesting to be the technical director. At the time, I had absolutely no idea what it was. Again, every six months, every year the tasks changed a little, but it was super interesting. It was interesting to get to the States as well, so I decided on it and just went there.
Vitaly Davydov: Let’s talk a little about the subscription. At first you had this free to try model, isn’t it how sneakers are delivered these days? They send the package to you by mail, and if they don’t fit or something, you send it back by mail as well, sometimes even without the courier?
Andrey Rebrov: Yes, texactly. I’m a very big fan of the Nike brand, but I don’t go shopping to clothing stores, I order everything to my home. I order several sizes, they come by post to my door, I try them on, and send back what I didn’t like. I just come to the post office, show the QR code, and they pack and send everything themselves. Taking into account the availability of credit cards, not only debit cards, this is very convenient.
Vitaly Davydov: How did you switch to the subscription model? Which is actually a subscription to some offline stuff, not a service subscription.
Andrey Rebrov: There were a number of obvious difficulties with the Try Before You Buy model. The first one was the unpredictable cash flow. The whole sending cycle was time-consuming: we send the fragrance to the customer, the package gets delivered by the courier or post, the customer spends some time testing it, and then sends it back. We were very dependent on the postal service and didn’t know how much the customer would buy. It was very difficult in the early stages. Especially because you invest a lot of money in the inventory. That is, you have a huge amount of inventory not yet sold at any given time. It’s either in stock, on the way, with the customer, or going back. And you don’t know for a very long time how much of it you’d be able to sell.
The second problem was fraud. We couldn’t immediately withdraw the entire amount from the person and even freeze it, because this would be no different from a purchase. We could withdraw only a small amount, but again there were no guarantees that we would then withdraw the entire amount. And it was a very big problem for our subscriptions as the number 1 reason for the recurring payments not going through was the lack of money in the client’s account. We have an unusually high percentage of debit cards used in our service, rather than credit cards. Although in the States, of course, credit cards dominate.
And the third problem was a very large check size. It was difficult for people to commit to a cost of $300 for three bottles of perfume, for example. Now our subscription costs $16.95, so even if you order three bottles, it will cost you less than $60.
This is basically the main idea of our service, that we reduce the risk of buying what you don’t like. What’s the problem with buying perfume online? If you have opened them, and you have to open them in order to try, you cannot return them. Because the seller wouldn’t be able to resell it. That’s why this check reduction, the human commit reduction, the transaction size reduction, make it more appealing to the customer.
Plus, this opportunity to try new perfumes month after month actually adds a certain surprise for a person and a certain willingness to stay with you. Because again, what I said at the beginning, thousands of new perfumes are released every year, and many of them are released by big brands, which naturally invest a lot in marketing. And people see these fragrances around, and they want to try them. Therefore, this desire to try something new remains. And our subscription very well allows us to monetize it.
Vitaly Davydov: Interesting. But how does it work exactly? I subscribe to a certain number of new bottles per month, for example, one. And every month you send me a small bottle, which is enough for about 20 days a month, for like, every working day?
Andrey Rebrov: Yes, and it’s usually enough for a whole month.
Vitaly Davydov: Should I return the old one or not?
Andrey Rebrov: No, you keep it. In the first month you receive a case and a bottle, and then every next month you receive a pure bottle of perfume that you put in this case. One of the key differences between our service and others is that the fragrances you receive aren’t random. You decide what you want to receive month after month. As a rule, people take the first two or three perfumes that are either super expensive in retail or are very well promoted. And only then they start experimenting.
Vitaly Davydov: How is money processing arranged in this case? In the mobile world it’s Apple, Google, at least Stripe. But how does it work for you?
Andrey Rebrov: It’s one of my favorite topics, because I was into it a lot at one time, and still continue to learn more as it’s actively developing. There are now a huge number of subscription platforms. As a rule, the same services are used for both the physical and digital worlds, but there may be some nuances associated with making payments and calculating certain metrics.
There is Recurly, Chargebee, and Zuora, to name a few. Stripe is also a subscription billing platform, and we actually used this one when we started. But let’s have a look at how the fintech stack of a subscription business looks like. There’s your online store. Then you have some kind of a subscription billing platform that is responsible for subscription plans, billing, error handling and dunning, which is the restoration of failed payments. Further goes the acquirer itself, which is called PSP (Payment Service Provider). These are ADM, Worldpay and so on. Further, respectively, the acquirer is followed by Visa, MasterCard and all card brands.
But in parallel with the traditional payment stack, there are all alternative payments like PayPal, Amazon Pay, Digital Valet, ApplePay, Google Valet and so on.
Vitaly Davydov: is it true that Dunning can be really critical on revenue? I heard some stories, I don’t know how true they are, that when people switched to Stripe, their revenue increased simply due to the fact that Stripe was able to charge at the right time, e.g., when salary comes.
Andrey Rebrov: Yes, it’s true. We used Stripe for a couple of years, but then changed two more payment systems. We left Stripe because we had a big problem with the payment restoration. It was possible to make a repeated payment only three times, but we wanted more. And we had to refine the logic for making additional payments. And when we increased their number, we were able to reduce the involuntary churn.
The second problem we had with Stripe was that they had a very poor error description of why the payment failed. They had about five errors, and 90% of the payment issues fell into the general decline category, that is, some general reason for the payment failure. It was impossible to work with this, because we couldn’t send people a specific letter of the reason for the failure of their payment, so that they could fix it. And this is crucial, because once we switched to Vindicia and were able to make personalized messages, the churn metrics changed significantly. All this has a significant effect on the payment process.
Andrey Rebrov: There’s also an aspect of automatic payment processing – and the services compete in this field with each other. It’s all about statistics: the larger the service, the more statistics it sees by region, by the bank that issued the card, and so on. For example, over the past few years, all these recurring payment platforms noticed that on Friday morning payments start to flow better. Why? Because a lot of people switched to the gig economy and began to receive payments once a week, on Friday. They also noticed that payments flow better on the 1st and 15th, because there’s a huge number of people whose salary is tied to the 1st and 15th. The worst time for payments is Monday morning, because people spent money over the weekend and on Monday they simply don’t have any.
Therefore, many companies have their own departments, not only data science, but also a data analyst who analyzes patterns, and prepares reports for a particular merchant, so that the merchant can somehow tweak and supplement the practice of making regular payments. We regularly do this as well, tweak some things based on the patterns that we notice.
Vitaly Davydov: Wow! This requires writing custom logic? That is, the subscription provider gives some kind of Low Level API or SDK that allows you to make a payment. But then you decide the rest?
Andrey Rebrov: It depends on the platform. One says: look, this is the behavior we can adjust for you. Another says: look, we basically have AI to do all this. And some other says: look, we have AI, but here’s also API for you, you can still tweak it to your liking. And once again, this is why it’s important to look at the error codes. Because for each error you can create its own logic.
But it’s important to note that you can’t recharge endlessly, because you pay for it. And if you do this a lot, Visa and MasterCard will notice your “strange behavior”, because all merchants have a unique ID. And if you make more than 15 attempts to make a payment, they’ll make you pay more from the 16th attempt. There are a lot of aspects of this entire payment logic, and you need to dive into it when you have changes. If you raise by 0.1%, for you it will be the result of tens and hundreds of thousands of dollars, and this is quite a lot.
Vitaly Davydov: That’s wild. In the mobile industry Apple takes care of all of this, we don’t even know that something like that’s happening behind the scenes. But do you have chargebacks as well?
Andrey Rebrov: In fact, yes, chargeback or dispute, as they are also called, is a problem for any business in the States. And I think more so for the physical ones. Because in a physical business, unlike a digital one, you are tied to logistics partners. These are all UP mail services, USPS, commercial UPS, DHL, FedEx. Whatever you use, you depend on them.
And for consumers there is no difference if it’s USPS or someone else. In their world picture there’s only the merchant. if something was not delivered – the seller is to blame. Something broke while being transported – the seller is to blame.
Chargeback is a situation when you, as a consumer, think that the seller or service provider has not provided you with the service, or did it incorrectly and violated the terms and conditions. You contact your bank through a mobile application or your personal account, choose the payment and say: “I’m making a chargeback.” You choose a reason and the bank returns the money to you and then the seller or the one who provides the services must provide the necessary documents to show that everything was done correctly.
For example, in a common scenario in e-commerce, a person says: “I didn’t receive the package, so I want to make a chargeback.” In this case, we track the entire path of the box and check all documents in the chain. If everything is fine with them and the carrier delivered the package, then we tell the bank that the client is lying. In such cases, the bank takes the side of the seller, and does not withdraw any money nor take the fine from the e-commerce.
Very often, chargebacks are not made for legitimate reasons. Very often they are fraudulent chargebacks. This is what is called friendly fraud, when people try to deceive the merchant. In some cases, for example, our transaction is not that big, 16.95, sometimes it’s cheaper for us to just accept it and block the customer permanently.
Vitaly Davydov: What’s better, refund or chargeback?
Andrey Rebrov: Refund, because with the chargeback, if you couldn’t prove it wrong, you end up paying a fee. And if you have a certain percentage of chargeback per month regarding your transactions, your network takes note of that and starts looking at you very carefully.
You must lower this percentage, because if you can’t cope, the cost of passing the chargeback increases from 15 to 150 dollars, and you start paying a monthly fine. After a certain number of months like that, your merchant ID gets banned.
Vitaly Davydov: Is it possible to make an attack on someone like that?
Andrey Rebrov: Yes, that’s what they do. We were once attacked by a botnet that just started to carry out fake transactions. We had to urgently and very seriously improve both anti-fraud tools and the security on the site in general.
Vitaly Davydov: We once talked about headless commerce with you. Tell us briefly what it is about, what is headless commerce in general?
Andrey Rebrov: When we started making Scentbird, one of the first questions that we had was whether we use some existing platform, or do it from scratch. And at that moment I insisted we make it from scratch, because we didn’t know what we were going to do. And then eventually we turned from the try before you buy, to the subscription model.
So our entire site, for the most part, is completely our own solution that we made ourselves. This is one of the ways to run an e-commerce or subscription business, make the platform yourself. We had to do this because we have a lot of unique features on the site. If it’s a standard e-commerce or with a little customization – people use the platform like Shopify, WooCommerce, and so on.
The problem is that when you have grown to a certain scale, you want to go beyond the standard functionality that the platform provides you. And, in fact, even when you are still a small player, you can still face the limitations of the platform. That is, if you want to do some A/B testing or calculate something differently, you cannot do this. So, when you’re big, you have to deal with personalization. You already have a very diverse audience. And you want to make some specific sales, or personal discounts and special offers.
For a long time there were two options on the market – custom development or a ready-made platform. But some time ago there appeared Headless Commerce, which takes the approach of microservice architecture. E-commerce by its nature is a set of certain functional blocks. You need to be able to add products and display them, make payments, work with the shopping cart and orders, collect, pack, manage, make refunds, and so on. So there are many functional blocks. When you are engaged in the custom development, you make these blocks yourself and integrate them yourself. When you use a platform, it’s all in one box, and you have limited possibilities to work with it.
Headless Commerce says: “Look, guys, we’ll provide you with each of these blocks separately. We’ll give you an API to it, and you’ll be able to use the blocks that you need.”. In E-commerce, you need a shopping cart, you need a showcase, you need certain types of discounts, you need to bring the product, place an order, and send it. So you can use the Headless Commerce platform for this. Due to the fact that this is a set of separate services, we can easily integrate this into the already existing ecosystem of Scentbird services.
We can configure anything on top and seamlessly integrate this with our subscription. It’s called Headless because there’s no frontend, so we can bring it anywhere to our interface. And once you’ve integrated the basic set of e-commerce functionality, you can add the services you need from other providers to solve certain problems. There are certain pros and cons to this approach, however. The cons are that you’ll need a permanent integrator partner who both supports the platform and makes new integrations. But then again, you have that flexibility.
And the advice that is often given is that you need to have your own team, which will be doing the maintenance in the future, and the integrator who will explain to you how to do the integration correctly and build it all with you.
Vitaly Davydov: To sum it up: there is some kind of engine that has an open API, and with the help of this engine you can collect different e-commerce components and “weld” them into something you particularly need.
Andrey Rebrov: That’s right.
Vitaly Davydov: Cool! But I thought other platforms worked the same.
Andrey Rebrov: In theory, yes. But when you use Shopify and connect something with a plugin, it works the way the creator of this plugin intended. You can’t tweak it and make it do things it cannot do by default. If there is no necessary setting, you’re just left with what it is. Sometimes you can make crutches in Shopify, but no one guarantees you that if a new update comes out, nothing will break.
Nikita Maidanov: But can you also use the card subscription and all these services in a mobile app, as you deliver physical goods?
Andrey Rebrov: Of course, we have our own mobile app for iOS and Android, and they work just the same with our APIs. People buy both subscription and e-commerce products. Again, as we sell a physical subscription, we aren’t required to do this through the Apple payments, and aren’t required to give the 30% commission. If you take the default Stripe pricing that you get at the start, the commission is 2.8% of the transaction size, plus 30 cents. Then, of course, with your growth, you can negotiate and reduce it. But you pay 3%, not 30%. The marginality in e-commerce is not so high because you work with a physical product, so every fraction of a percent is of great importance.
Nikita Maidanov: Where do the users onboard better, on the web or on the mobile?
Andrey Rebrov: It’s hard to say. We see that people tend to get to know the service through the web version, but the mobile web version has more traffic. In terms of purchases, it’s 50/50 between desktop and web. And mobile apps are very cool for helping with LTV improvements and repeat purchases. It’s more convenient for people to manage their subscription in the native application, because it gives a more pleasant experience.
Nikita Maidanov: That’s interesting. Maybe you can give some advice to our listeners who are just getting into all this and thinking about integrating web payments?
Andrey Rebrov: For those who use web payments, I would first of all say that when you start, there is no need to reinvent the wheel. We started with Stripe, and it gave us a good opportunity to launch and focus on the business further. Start with what is comfortable, especially if you don’t have some crazy customized subscriptions. You need to prove that your business model works and that your product is of interest to someone first. So use what works well, what is time-tested. When you grow up, you’ll realize that you’re missing something, and you will start to change.
Nikita Maidanov: Thank you for coming, it was very interesting!
Andrey Rebrov: Thank you for inviting me!
September 6, 2022
13 min read
September 8, 2022
5 min read
December 4, 2020
12 min read