A look under Stripe’s hood: Using data to demystify payments
分组讨论, Payments
运行时间
填写表单,观看完整视频
Great payments experiences drive conversion, but figuring out the best approach is increasingly difficult as customer expectations evolve, fraud increases, and payment methods proliferate. See how Stripe’s data science and product teams use A/B testing and experimentation to help you improve your payments experience. We also answer the top payments questions we hear from Stripe users, and speak with businesses that are delivering best-in-class payments experiences.
Speakers
Carlyle Nicoll, Data Scientist, Stripe
Théo Blochet, Product Manager, Stripe
Hilary Platt, Account Executive, Stripe
Jonathan Cordeau, VP Product, Payments and Globalization, Peek
Brian McMullin, VP of Product, SamCart
CARLYLE NICOLL: Hi, my name is Carlyle Nicoll and I'm a data scientist at Stripe who works on payments. One of the top asks from all of you has been more first-party insights on how Stripe thinks about our payments experience. You, understandably, want to leverage these insights to improve your own buyer experience. Today, over the next 30 minutes, we're going to look under the hood at a couple of experiments to show you how Stripe thinks about causality. As a data scientist who's personally worked on many of these experiments, I'm so excited to dig into the data with all of you.
But before we get to the data, I want to underscore the strategic importance of getting payments right. In an ever-expanding global economy, the most important growth and innovation opportunities depend on payments and improving that for our customers. It's obvious to all of us what a poor payment experience looks like: endless fields, sketchy card forms, or only accepting an obscure cryptocurrency. You know this because you've done the hard work of acquiring customers in the first place and getting them to your site. Yet, with all this information, it's still really difficult to discern exactly which payments investments matter most to your business. Getting the data to make strategic payments decisions means you have to have many things, enough resources, enough traffic, and the ability to run end-to-end tests that measure more than just click-through rates. You also have to factor in the cost of processing, the cost of refunds, and the cost of disputes, all to measure your bottom-line impact.
And this is why we're here. Stripe processes more than $1 trillion dollars in annual revenue. This is 1% of the annual GDP. This gives us a wealth of data to run powerful experiments with the singular goal of making payments easier for you and delivering a better experience for your customers. The interplay between outcomes is nuanced in all cases, purchase clicks are up, but auth rates are down, or conversion rates are up, but fraud rates are too. What businesses is that good for? Should Stripe shift or change our features?
We are constantly running A/B tests, the gold standard in causal inference, to measure all of this for you. And we also dive into observational data when an experiment is impossible. Today we're going to look under our hood at a few experiments to show you, in practice, how data science helps you optimize your payments. We'll answer your top questions about express checkouts, about buy now, pay later payment methods, and about buyer localization. And buckle up, the data just might surprise you. Let's start with express checkout options. We know checkouts need to be fast, but how fast? We took to the streets and asked people how long they expect a checkout to take. Here's what they had to say.
[VIDEO]
A seamless checkout will usually take literally less than 30 seconds to pay.
Less than 30 seconds.
10 seconds.
10 seconds.
Two seconds. Maybe I'm too ambitious.
CARLYLE NICOLL: Pretty ambitious, because less than 10 seconds is pretty fast. It's no secret that making your online checkout experience fast is absolutely critical. I'd like to bring out a member of our product team to tell us a little bit more about the online checkout experience. Please welcome Théo, product lead for payment methods.
THÉO BLOCHET: Hi, Carlyle.
CARLYLE NICOLL: Hey, Théo. I was just telling everyone how making your online payments fast is absolutely critical.
THÉO BLOCHET: Yeah, there are many things that add up to a less efficient checkout. Things like lack of autofill or not letting repeat customers reuse their payment method. Having buyers go through too many web redirects to authenticate a payment. The list really goes on. If I had to pick one recommendation though to improve speed and conversion at checkout, it would be to skip it.
CARLYLE NICOLL: I'm sorry, did he just say to skip the checkout?
THÉO BLOCHET: In a way. So think about the times when you're looking for the latest viral product, you found it, and then right then and there, there was a Google Pay or Apple Pay button for you to just get the purchase over with.
CARLYLE NICOLL: Honestly, I did just buy another mini waffle maker with Apple Pay this weekend, with no hesitation..
THÉO BLOCHET: So it's like a waffle maker, but like a mini version?
CARLYLE NICOLL: A mini waffle maker. I saw it on Waffle Tok. Is anybody else making chaffles? I see one guy over there. He gets me. He gets me.
THÉO BLOCHET: Not me. Not me.
CARLYLE NICOLL: Anyway. You're saying that express checkout options are a way for people to make faster payments with Apple Pay and Google Pay. Is there anything else?
THÉO BLOCHET: Yeah, so those payment methods, like in the case of your waffle maker make it possible to really expedite the checkout process before or at the time of checkout. We've actually looked under the hood at aggregated data from businesses on Stripe. Through a thorough A/B test, we checked what the difference in conversion was when offering Apple Pay at checkout and not offering Apple Pay at checkout. What we saw was a 10%, statistically-significant conversion uplift from adding Apple Pay. That's just one express checkout option driving a double-digit relative conversion uplift. When we look at speed, when we compare the speed of checkout sessions made using Apple Pay, Google Pay, or Link, Stripe's accelerated checkout solution, we observed a three-times faster median checkout speed with those methods compared to other payment methods on Stripe. That's significant. Actually, Carlyle, didn't data science just run an experiment on Link?
CARLYLE NICOLL: We did. We just ran an A/B test on Link among businesses with a high repeat customer base. In the treatment we showed Link, and in the control we did not, which was the standard experience. Here's a Link experience if you didn't already know. Link allows customers to save their payment details and autofill checkout forms for future purchases on Link-enabled websites. When Link was offered, these businesses saw an average returning user conversion uplift of 14%. That is huge. Even businesses with returning customer bases were seeing a double-digit increase in conversion. People who already had high intent and were already really excited, they were able to spend their money faster. It's from this data that we could reasonably infer that Link was providing a faster checkout with higher conversion rates that translates to more revenue for your business.
THÉO BLOCHET: So higher revenue uplift, higher conversion rate, checkout speed. If you needed any more of a reason to go implement those express checkout methods at checkout, we offer something called the Express Checkout Element. It's a single embeddable element that gives you one integration to offer all of those one-click checkout buttons, and dynamically shows the right one-click checkout button to the right buyer, with more coming soon.
CARLYLE NICOLL: Great, so we all can buy more mini waffle makers just as much as me and that one guy in the back do?
THÉO BLOCHET: Yeah, or all other things.
CARLYLE NICOLL: Okay.
THÉO BLOCHET: So before you run out the door to explore more viral waffle recipes or add more express-checkout options to your checkout, there's more that you can do to try and improve conversion at checkout.
CARLYLE NICOLL: Amazing. Well, I'll leave you to share more with the audience. I have to go set up a couple more experiments.
THÉO BLOCHET: All right, see you Carlyle.
So. I traveled here from Berlin, Germany, just getting over the jet lag. In Germany and in many European countries, buy now, pay later methods are extremely popular—so popular, in fact, that 28% of ecommerce volume in Germany goes through buy now, pay later methods. You may be thinking, “Well, I don't sell in Germany, so this is all nice, Théo, but not for me.” Or you may be saying buy now, pay later methods are not for my business, they are for businesses with high order value, expensive services, or expensive goods. So I actually looked into the Stripe data to try and answer that question. What is the median checkout amount for transactions made using buy now, pay laters? Intuitively, I would've thought, before I saw the data, that it was around the $100 range, $150 maybe. When our data science team looked into the data, they found that a median amount for buy now, pay later transactions is actually $27. Meaning 50% of transactions by account, across the Stripe network, are in that $0 to $30 range, which is quite astounding.
What we can then conclude from that when we look at conversion impact is that businesses across the order-value spectrum actually get a meaningful conversion uplift from adding buy now, pay later methods. Of course, if you have a higher order value, you have a higher uplift, but if you look across the range, the average is a 10% relative-conversion uplift when you add buy now, pay later methods. You may be wondering which buy now, pay later method? The answer is, TL;DR: all of them. (Théo chuckles) They have, broadly, mutually exclusive user bases. So when you use them together, you actually get better coverage of your buyer base, so a higher revenue uplift.
Now, you may be asking, “Buy now, pay later methods, this is all great, but I'm already offering cards, I'm potentially offering express checkout options, I'm offering wallets… won't adding buy now, pay later methods just take my existing payment volume and just charge me more for it, because those methods are more expensive?” This cannibalization concern, that's how we call it, is a very valid concern and that's why many of you haven't added buy now, pay later methods to your checkout just yet. So again, we teamed up with our data science team and we ran an experiment. Across 150,000 checkout sessions, we ran an experiment where we either removed or added buy now, pay later options at checkout, and then we looked at the impact on the payment method mix. We ran the experiment across varied business types, across different buy now, pay later methods like Affirm, Afterpay, Klarna, [and] a combination of those. The result varied based on the experiment, but what we found is that 61%—that's like the most conservative estimate of all of those experiments—61% of the volume was incremental, meaning those are customers or revenue that otherwise would just have been lost at checkout. So largely, buy now, pay later methods do not cannibalize your existing volume. They add to it.
When you look at the overall conversion impact for buy now, pay later methods, what we found in our experiments is an 11% revenue uplift. Part of that is the 10% conversion uplift, which was talked about. And the other part is higher cart value, higher average order value. This more than offsets the light cannibalization of those methods and their higher transaction costs. So we've covered some data points from the Stripe network of businesses around what quantified impact you can expect from express checkout methods, from buy now, pay laters. But there's more, I'd like to invite Hilary to tell us about localization.
Hi, Hilary.
HILARY PLATT: Hi, Théo.
THÉO BLOCHET: So Hilary, you work day in and day out with Stripe customers to help them make the most out of Stripe's suite of checkout products. What more can you share?
HILARY PLATT: Yeah, so I heard you talking about payment methods, and I recently relocated from San Francisco to London and I have some interesting insights to share about the importance of localization.
THÉO BLOCHET: Awesome, I'll let you take it from here.
HILARY PLATT: Thanks, Théo. All right, so one of the unexpected eye-openers for me in moving abroad, besides the fact that people do in fact say “jolly good,” is the plethora of payment methods that everybody is using. My multinational colleagues, they have all these different ways to pay, and I just have my one credit card. I mean, I do love my points. And when I've traveled abroad to other parts of Europe, payment methods like iDeal, Blink, Bancontact, these are all household names. I tell you this to illustrate the point that location has a significant impact on customer preference. But how big of a deal really is it to offer these local payment methods? Aren't the globally supported payment methods that Théo talked about enough? The answer is they're quite significant. Very important. 85% of customers abandon their cart if they do not see their preferred payment method at checkout. 85%.
I honestly find this stat a little bit tragic because it means you as a business have gone all the way to make this amazing product that customers have put into their cart, they've gotten to the checkout page, and they've abandoned it just before they clicked buy. All could be solved. And sadly, many of these customers probably won't come back to your page knowing that you don't offer their preferred payment method. So let's take it back to the streets of New York City to see how people like to pay.
[VIDEO]
Apple Pay.
PayPal. I use Klarna.
Klarna and Afterpay.
Wish.
My credit cards.
Paytm and Google Pay.
... or Shopee.
I usually try to use whatever is the fastest way to pay and get on with my life.
HILARY PLATT: I'm very much so with that guy, let's get on with our lives. But you as a business, you might be thinking to yourselves like, “Okay, there's all these different payment methods, how do I know which one is best for my business?” This is where Stripe's scale and investment in data science is a huge advantage. We're able to give you real, first-party data on your customers. Specifically, what payment methods do they like to use? For example, we found that when businesses on Stripe add a popular local payment method, that payment method becomes the preferred payment method at checkout. So you see here across the Netherlands, Poland, Malaysia, Brazil, when we offered these individual payment methods, they became the preferred choice.
So one thing that I haven't mentioned yet is they become the preferred choice, and this is a lot for you to think about. You have to think about integration, maintenance, it's a lot of work. That is why Stripe has built our optimized services to allow businesses to easily integrate new payment methods and our dynamic payment methods to surface the right payment method to your customers based on their buyer location and the transaction amount, browser and currency. Currency is an interesting one because a lot of these popular local payment methods like Swish, PayTo, OXXO, they only work if you have that local currency presented.
You might be thinking, “Okay, this is great, but my customers are all in one region. I don't need to be thinking yet about these local payment methods from other countries or local currencies.” Totally possible, you know your customers best. But I do want to share an example of how we worked with an emerging AI company to make a pretty important payments discovery. So a little bit of background on the business, US business, very focused on North America. They were very surprised to learn that actually 50% of their transactions were coming from international buyers. Brazil was their second largest market. And so, since they were so US focused, they were only presenting in USD, and this ended up having a really significant impact on their bottom line, starting with conversion. So when a US customer landed on their site, there was a 51% conversion rate, which meant the customer gets to the page, they click buy. 51% in the US.
In these European countries, [as] you can see here, significantly less. Brazil, second largest market, 17%. What's more was their authorization rates. So again, this is a customer actually [pressing] buy. What percent of those transactions were actually accepted by the card networks? US, 90%, Brazil, 55%, global mix, the remaining countries, 68%. Just imagine what this revenue uplift could have looked like if they'd been presenting in a local currency or offering a local payment method. Well, thank you to Carlyle and her team, data science, we have some more data to share.
So we ran an experiment coming out of the UK, my new home country, and this was a holdback experiment using our adaptive pricing product. So this automatically presents customers with options to pay in their local currency. So for example, on the left here, we didn't offer adaptive pricing. This meant that European customers were only able to check out with pounds. On the other side of the screen, we were using our adaptive pricing product, which meant that customers had the option to pay with either euros or with pounds.
The results: we found that businesses using Adaptive Pricing saw a 17.8% relative uplift in their international revenue. This is huge. This will grow in importance for our existing customers as they're looking to go global, and then for new businesses that want to be global from day one. As we like to say at Stripe, being global means being hyper-local. So, I would like to invite to the stage two of our customers to hear how this plays out for them. Please invite Brian McMullin from SamCart and Jonathan Cordeau from Peek Travel to join me on stage.
All right. Thank you for being here. Before we get started, I'd love if you could just each introduce yourselves, your roles, and a little bit more about your businesses. Let's start with you, Jonathan.
JONATHAN CORDEAU: So we're amongst a couple hundred of our closest friends, so my friends call me JC, you can call me JC. I'm the VP of product for payments and international at Peek Travel. Peek's on a mission to connect the world through experiences. And so if you think about after the event tonight, if you wanted to go up to North Beach and do a walking food tour or maybe rent some paddleboards to do some yoga, those businesses use Peek as their operating system. So I like to kind of joke that my job is to make sure that thousands of those kinds of businesses can collect billions of dollars around the world.
HILARY PLATT: Awesome. I always rent a paddleboard when I want to do yoga, so thank you, JC.
BRIAN MCMULLIN: All right, great, I'm Brian McMullin, VP of product at SamCart. SamCart is an ecommerce platform for creators. We help them take their passions and bring them online with super fast and powerful checkout experiences. So think about the yoga studio that we talked about this morning, we heard about this morning, bringing their sessions online, or a vegan chef, you know, creating online courses and selling recipe books. Or a health and wellness professional, bringing custom keto programs online, selling alongside protein supplements. So the way I like to think about my job and what we do is we help creators take their side hustles, turn them into main hustles, and then ultimately help turn them into thriving growing businesses
HILARY PLATT: Awesome. Thank you both. Well, so we've talked a lot today about customer preferences and businesses needing to adapt to those preferences. I'd love to hear how you all at your businesses are thinking about adapting and what has had the greatest impact. Back to you, JC.
JONATHAN CORDEAU: Yeah. So you can't really connect the world through experiences if people can't pay for those experiences, so this is a pretty critical part of our business. We spend a lot of time looking at the entirety of the booking flow, and obviously the payments page being a really big part of that. If we look at a customer like the Museum of Ice Cream in New York, so you have a lot of local traffic, but you have a lot of international traffic as well. And so if you think about your experience when you're in another city, you're traveling, there's two things that are really important that we can look at the data to tell us that it really drives this kind of behavior. Number one, the speed to checkout. What we see is you're, again, you're in New York, you see an ad for the Museum of Ice Cream, it becomes a little bit of maybe a, you know, last minute decision that you want to make.
And so being able to move through that checkout experience quickly is incredibly important. Very competitive environment. The second thing is, we looked at the data and see that over 70 of our bookings are actually made on mobile, which doesn't seem like a big surprise, but is a really powerful piece of information that you can use to then drive the decisions that we make. And so both of those areas really tell us that they can be served well by wallets. And so about a year ago, we made a decision to move forward with Apple Pay and Google Pay. And the decision paid off. So we get to kind of reinforce some of the things that you've been talking about. Apple Pay increased conversion by 7.8%, driving real revenue back to the bottom line for our partners. It also increased the speed of conversion by 15%. So real drivers behind those customer demands that we, you know, saw in the market. That then drove us to then implement Link as well, to see.
And what's really surprising about Link, we've been talking about [it] a lot, but there's a segment of the market that might not have a card enabled in their Apple Wallet or their Google Pay. And what we saw is that segment of the market, they actually now were given a wallet that they didn't have that option before. And so we saw [an] additional increase in that market specifically. And all of that has really driven us to continue to move down this path of additional payment methods. We've talked about iDeal, we implemented iDeal back in Q4 and saw an immediate share shift of 30% of all payment volume moving overnight to iDeal, which really speaks to the pent-up demand that customers had that wanted that payment method.
HILARY PLATT: Awesome. For those of you who don't know, iDeal is for the Netherlands. So if you're operating there and don't have it today, hear that, 30%. Who here has been to the Museum of Ice Cream? Did you check out with Apple Pay or Google Pay on Peek? Let's hope so. All right, Brian.
BRIAN MCMULLIN: Yeah. So, as an ecommerce platform, you know, check out and payment experience is critical for us. So very similar to JC, digital wallets, Stripe Link, buy now, pay later, mobile optimization, has been key for us and making sure that we're driving increased conversion rates and overall AOV per transaction. Although where we've seen some of the most exciting, you know, impact is in our experiments with embedded checkout and really thinking about how you can decrease the space between the buying decision and the ultimate purchase process. And it turns out that the highest-converting funnel is actually no funnel at all. And being able to take that checkout experience and place it where you're engaging with your customer about that product significantly increases your chance of completing that purchase. And we've tried this ourselves, we've used it, we use it today on our own pricing page, selling SamCart, and we've seen an increase of upwards of 48% in conversion rate just by having a button on that pricing page where people can buy right there.
HILARY PLATT: That's awesome. It's clear to me that you both have very good data science teams, which we love to see coming from Stripe. I'm curious, what is top of mind for both of you in the next 12 months? Where are you putting your investments, Brian, maybe?
BRIAN MCMULLIN: Yeah, so in the payment space, I think two big things are coming up for us. One, continuing to optimize our checkout and payment experience. And what that means for us is making sure that we are offering all the payment methods that our customers and their buyers want. So on one hand, I think JC has firmly sold me on iDeal at this point, but we're hearing a lot from our customers that they want to push into Europe and have local payment methods there, so iDeal, SEPA, you know, things along that line are coming up for us. I think on the other side of payment methods, a lot of interest in bank payment methods, the lower-fee payment methods, so ACH Direct Debit, there.
And then I think more on the kind of platform and operational side, tax and reporting has come up as a really big interest and something that our customers are looking for within our platform. As you can imagine, people come to our platform for all the flashy, great, you know, checkout experiences and conversion rates, but they ultimately stay because we are a great platform for them to grow their business and being able to support a lot of different tax capability, and then, ultimately, report on their success as a business is critical for them. So we're going to be spending a fair amount of time there partnering with Stripe to roll out Stripe Tax later on this year.
HILARY PLATT: I feel like I've never had more exciting customer conversations when we tell them that we do tax, and they're like, “Oh my God, we get to get rid of this.” So, excited to hear you're having those conversations. JC.
JONATHAN CORDEAU: Yeah, top of mind for everyone I know is growing revenue, and so we work with thousands of these small, midsize enterprise businesses that we need to think about how can we help them drive more money on their top line and then also does it help us maybe increase our margin as well. So that that's kind of driving everything we're thinking about from a roadmap standpoint. When we look at just excitement within that partner base, we can see buy now, pay later is something that we'll be looking at.
Obviously with the success that we've had with the wallets and iDeal, we're really looking at additional payment methods, as well, as a way to drive increased conversion, maybe reduce some of those costs as well. And then just the theme that we've been seeing along the way is how do we continue to experiment because we control the entire booking flow, not just the payment page. There [are] a lot of little micro-optimizations that we can look at and test over the course of time. We do a lot of transaction volume and that'll be something that we'll really dig into further over the course of the next few quarters to find just those little ways to shave the basis points in the right direction.
HILARY PLATT: Awesome. All right, I'll finish with the age-old question, which is, what is the advice that you could give to these product and payments leaders in the audience to see some of the same success that you both have seen? So Brian.
BRIAN MCMULLIN: All right, so I have two things and I swear that I came up with these before the keynote this morning, and I'm not just taking that and running with it. But I'd say the first one is make sure you're treating your checkout and your payment experience like a core customer experience. Which, to me means understanding how your buyers really want to buy. We almost have too many options for us to choose from right now between alternative payment methods, digital wallets, buy now, pay later, local payment methods, Stripe Link. It's kind of overwhelming, not only for us, but obviously for buyers too, and the last thing that we want is our checkout experience to look like the Cheesecake Factory menu where it's like, so many things that you get overwhelmed and you don't know what to order. So thinking about and understanding how your buyers want to buy and making sure that you're really focused there, and leaning on Stripe and their capabilities to help you optimize that really out of the box.
I think the second thing is, and we were kind of laughing about this earlier, but, beware of the silent killers of payments. You know, really being things like fraud, bad card auth rates, and ineffective dunning. These are the types of things that, if left unchecked, can really erode revenue, customer retention rates, things along that line. And, if we're being honest with ourselves, ignoring that and not investing there is kind of like skipping leg day at the gym. You can have the best payment and checkout capability on the front end, but if fraud is running rampant or banks aren't accepting your payments, you really don't have a holistic, healthy business. So thinking about adopting things like Adaptive Acceptance, Smart Retries for dunning, and Radar so you can hopefully block fraud before it even happens by looking at what card types are working well for you, what bins you should be avoiding, etcetera, and spending a little bit of time there will make sure you have that full financial health beach bod ready to go for your business.
HILARY PLATT: The leg day avoiding fraud. It's good. We'll get that into the marketing somehow. All right, Brian, or sorry, JC.
JONATHAN CORDEAU: I'll go, I know we're up on time. I think I'll just add two more things, I couldn't agree more. If you're not, we're talking a lot about front-end changes, if you're not thinking about the backend, if you don't know L2, L3 data, the data that you send along with the transaction can be significantly helpful in increasing authorization rate, whilst decreasing cost. So definitely look up L2, L3 data. And maybe just a reminder for all of us, we made some assumptions, we talked about some of the things that we maybe thought were going to happen. And so, test those assumptions.
Don't be afraid to experiment, don't be afraid to go out and get good tools if you don't have amazing data science teams within your organization. And talk to customers. Yeah, these are great videos that just reinforce the fact that sometimes it's fun, you get some good quips, but there's also, the market is evolving continuously. And so, we know that, we know we're supposed to talk to customers, especially the product folks in the audience, but it's a good reminder that they hold that information that we really can use to then drive making good decisions to deliver those outcomes.
HILARY PLATT: All right, well please give a round of applause for JC and Brian. Thank you.
All right, well we hope you've enjoyed this look under the Stripe hood to see how we are using experimentation and data to demystify payments. Our goal is really to give you a clear ROI when it comes to making improvements to your checkout, whether that is adding express checkout options, integrating buy now, pay laters, localizing with the right payment methods and currencies, or hopefully all of the above. The bottom line is that we want to help you increase your bottom line.
So this is why we are backing our products with Stripe-run experiments and we are giving you the tools to run your own experiments with payment method A/B testing, fraud insights, and more. Please stick around for our payments roadmap breakout starting in 15 minutes, where we'll be talking about our payments investments. And please drop by our demo booth downstairs, to see our products in action. Thank you everyone for joining.