More recent writing on PayPal:
PayPal January 2024: Going Beyond Payments and Becoming a Commerce Platform and The real threat to PayPal and Stripe: Shopify
Online Payments in 2023
The last 8-10 months were fraught for online payment processors:
Adyen N.V. (ENXTAM:ADYEN) announced H1 earnings in August and dropped nearly 50% on news of compressed margins and weaker-than-expected growth in the US.
Stripe raised a $6.5 billion down round, cutting its valuation from $95 billion to $50 in March. Their IPO timeframe remains uncertain.
PayPal (NasdaqGS:PYPL) PyPayl grew net revenue by 15-20% YOY in FYs 15-21, but growth slowed to 8.5% in F22. Similar growth is expected in F23. High-margin processing has been under pressure from ApplePay and a decrease in consumer discretionary online spending post-COVID. Margins are suffering as growth becomes dominated by lower-margin unbranded processing with Braintree.
This outcome was predictable if you believed that, when stripped down, online payment processing is undifferentiated and nearly commoditized in the US market. Many argue that online payment processing is a race to the bottom, especially for enterprise-focused processors like Ayden and Braintree competing for slimmer and slimmer margins. While core card processing tends to be undifferentiated, PayPal, Adyen, and Stripe are each unique in the way they address the very difficult issues ancillary to core processing, which we’ll call “edge cases”. These could be checkout conversion, fraud, multi-currency support, mobile wallets, checkout flows, or any number of other things that make payments difficult. To borrow from Simon Taylor, payments are easy, but edge cases are hard.
Ayden is vertically integrated as an acquirer, gateway, and bank, leading to higher margins. They have the most advanced handling of multi-currency and cross-border processing which is especially important in Europe.1
Stripe is developer and SMB-focused with the best of the best edge case handling out of the box.
PayPal is the only player to own both a processing layer (Braintree and the PayPal button) and a direct consumer end-user network (classic PayPal, Venmo, etc).
Square/Block is truly omnichannel, blending P2P (CashApp), brick-and-mortar, and online payments.
To be clear, there are many other players in payments, including direct competitors to Ayden, Stripe, and Paypal, but they tend to be really big incumbents with diversified portfolios. We’re narrowing the ecosystem to Ayden, Paypal, and Stripe because they were the first players founded for the express purpose of solving online payments, and their business landscape is evolving far quicker.
Investors dinged PayPal at the end of 2021 as e-commerce volume slowed and it became clear that lower-margin Braintree processing would drive growth. Analyst sentiment remains relatively positive with a majority recommending buy or better. The share price is hovering on the low end of the consensus price target.
Now, PayPal must figure out how to monetize an uncommon collection of assets — a processor, a digital wallet network, and a P2P payments network— to revive growth and resist structural margin compression in high-volume payments. Here, we’ll focus on PayPal and what it means to transition from being a high-margin, proprietary network, to an all-purpose processor facing increasing competition.
Branded vs Unbranded processing and declining margins
When PayPal started in 1998, few appreciated that building a P2P payment network would eventually drive high-margin, commercial payment volume in traditional e-commerce. What eventually became obvious, though, is that users would much rather click a single button linked to a “wallet” to check out on a website rather than enter credit card numbers. Many even pay directly from the balances in their PayPal account.
The PayPal button wasn’t only unique because of the “digital wallet” experience. In the US, it printed higher margins and demanded higher merchant fees than almost any other processor in the industry. The PayPal button could do this because:
Merchants were willing to trade the higher payment conversion for higher processing fees (3.5% + fixed)
As a closed-loop system, PayPal’s network eliminates the middlemen that come with processing card transactions. Account-to-account transfer costs PayPal nothing. Even when funding a transaction via a linked card instead of a PayPal balance, the majority of PayPal funding is debit-based or bank-based (cheaper), and PayPal negotiates card network processing (Visa, Mastercard, etc) at scale on the backend.
As Web 2 payment processing tech evolved (e.g. Stripe, embeddable payment forms), so did the single-click enablement of different payment options. This created a distribution advantage where processors would place the button on their own checkout UIs, and PayPal would get a cut as the network when users paid with the button. Until ApplePay launched in 2014, there was little competition for digital wallets at checkout.
ApplePay enjoyed the same distribution advantage, though. To the consumer, ApplePay and PayPal are the same thing: digital wallets. To the merchant, ApplePay involves lower fees because it runs through the card networks. To Apple, it had free distribution on 75 million devices.
Today, ApplePay and other branded payment buttons are rapidly eating the PayPal button’s market share of one-click online payments (others include CashApp, Google Pay, Shopify’s Shop Pay, and Stripe’s “Link”). This pressure has led to a substantial volume and revenue decline in PayPal’s dominant and highest-margin segment. The prevailing conversation is no longer about how to bring the PayPal button back to glory, but rather about what will fill the hole.
Braintree
Paypal purchased the online payment process Braintree in 2013 for $800 million. Braintree is a traditional online processor that allows websites to accept credit and debit cards. Unlike paying with the PayPal button (“branded processing”), customers don’t know whether their payment is being processed by Braintree, Stripe, or someone else when paying via card numbers. Paypal considers this “unbranded processing”2.
Compared to the PayPal button, Braintree is a drastically lower-margin product because it’s subject to standard card network interchange rates on every transaction. Merchants are also willing to pay less for a vanilla card processing UX. So not only does Braintree card processing cost 90 fewer basis points than branded processing, but it also makes a smaller margin after the other players in the processing value chain take a cut. In practice, the biggest drivers of Braintree’s volume wield significant pricing power and pay far less than 2.59%.
PayPal is growing unbranded volumes rapidly with notable clients including Uber, Booking.com, Stuhub, and OpenTable. Over 90% of Braintree's volume is focused on large enterprises (LE) that sign multi-year contracts and require substantial integration resources. While PayPal doesn’t report Braintree revenue explicitly, we can infer $400-410B (30% TPV) of Braintree volume in F22, and F23 growth accelerated to 40% in Q1 and 30% in Q23
Company-wide F23 net revenue and TPV growth are expected to stabilize at 8% and 12% respectively. This would signal at least a temporary stabilization in margin compression as both branded and unbranded growth stabilize and branded checkout modestly re-accelerates. The biggest takeaway, though, is margins are unlikely to return to pre-2020 levels, and forward growth is now predominantly a story of unbranded processing.
Unbranded processing is now outgrowing branded volume by 5-6x, and that trajectory is unlikely to change. This is a good outcome relative to the decline in branded volume, but it means that overall margins are in structural decline under PayPal’s current business model.
In response to lower unbranded margins, management now echoes a vision where Braintree will be able to leverage the PayPal network to sell higher-margin, value-added services once it has cemented itself as a player with LEs. With 30-40% YOY growth, PayPal is an amassing unbranded waterfront property4 with a lower-margin product that may be able to tack on high-margin value-add in the future.
The narrative around value-added services is harder to conceptualize with certainty: what services would actually penetrate into LEs? When and would their margin make a meaningful difference against payment revenue? Are these value-added services scalable from client to client (these builds, as in the example of Uber, seem to be custom and drawn out)? Value-added services have actually declined as a percent of net revenue YOY.
Note: Quality of Revenue
Industry-wide, large enterprises come with increased churn risk over the long term as they integrate multiple unbranded processors for redundancy and leverage. LEs tend to have the manpower and resources to execute a multi-partner strategy like this. At scale, competitive processing rates and redundancy can outweigh the cost of engineering resources to implement.
A positive secondary effect of PayPal’s value-added services is the reduction in churn risk. For example, at Uber, once fraud and chargeback handling, driver payouts, wallets, and PayPal / Venmo consumer benefits are tied in, it becomes less trivial to ditch (or threaten to) PayPal.
Moving Braintree Downmarket
PayPal in response to an analyst question about the race to the bottom:
[Large enterprise is] only a subsegment of our business. And so we have this immense white space on the SMB side, where that profit pool is actually very meaningful, we essentially are going from a position where we -- our biggest PSP business today in the chunky -- like 90% of our unbranded volumes are coming on the [large enterprise] side, at least predominantly in the U.S.
— Gabrielle Rabinovitch, September 2023, UBS 2023 Fintech Leaders Conference
This response is notable because she didn’t mention higher margin, value-added services, which has been the standard response to margin compression thus far.
Chasing downmarket volume with Braintree could be a more attractive solution to declining margins at the enterprise level because SMB merchants don’t have the same pricing power as LEs and they typically yield higher margins on the same core processing. There is a lot of higher-margin volume to capture downmarket, but the product required to serve SMBs differs starkly from LEs.
Capturing existing payment volume
Large enterprises process volumes in the $ tens of billions (e.g. Uber) and have deep engineering resources. Winning contracts at this scale can be likened to Microsoft winning contracts to be secondary cloud providers to LEs already using AWS. Clients at this scale have the strategic and resource capacity to diversify infrastructure providers or switch them out altogether to improve unit economies. It explains the feasibility of Braintree’s growth.
Conversely, SMBs down the market don’t have the luxury of spending engineering hours to diversify or switch payment providers, even for a few hundred basis point savings. This existing volume is locked in with its existing processor, and PayPal won’t poach it, even competing on price.
The exception is Paypal PPCP which allows merchants using platforms like Shopify to “plug in” their PayPal account with a few clicks. If PPCP picks up, the outlook on capturing existing volume could change.
Stripe vs PayPal
Management acknowledges that Braintree in its original form wasn’t built to support downmarket customers at scale. The company launched SMB-focused “PayPal Checkout” and “PayPal PPCP” in April of this year to bring traditional card processing to SMBs and low-code use cases with Braintree powering card payments on the backend.
In this way, PayPal itself may edge on cannibalizing existing branded payment volume by providing existing downmarket merchants with more flexibility to accept Apple Pay and cards.
The launch of PayPal checkout places the company in direct contention with Stripe. While the downmarket is big, the competition is more complicated. If PayPal is said to specialize in consumer networks and enterprise-scale payments, Stripe specializes in courting developers, startups, and the long tail of online SMB payment use cases.
Product considerations change when serving the long tail of clients. Stripe is arguably much better at all of these things:
A greater focus on developer experience and sentiment
Long-tail workflow and functionality customization matters
Value-added services matter more, especially integrations out of the box
Long Tail of Edge Cases
Building for the long tail is expensive. Engineering resources are precious, so SMBs and startups are likely to choose the processor that fits their workflow needs above all else. This raises the importance of out-of-the-box edge case handling and developer experience. Stripe focused on this from day one, supporting:
Fully customizable subscription management, customizable payment links
Invoicing and billing - automatic reconciliation
Fraud and predictive risk management (“Radar”)
Many others including company formation and card issuing
On paper, PayPal checkout supports the banner use cases today, but it will take time to compete with Stripe on the long tail of edge cases. Anecdotally, it’s clear from spending five minutes in Stripe that they’ve thought about almost everything. That took a very long time and a lot of focus to build – Stripe was founded in 2010. This is where Stripe provides real value beyond mere undifferentiated processing.
With Braintree as the processing backend, PayPal is in still a better position than most to execute on SMBs. And while few can match the goodwill Stripe stokes with developers, building client-facing APIs and embeddable components is nothing new to PayPal. Moving down the market will come at a cost, and it will be a head-to-head fight against a Silicon Valley juggernaut.
Value of the network
PayPal’s killer advantage as a payments company continues to be its network. No other payment company except Square (with CashApp) can come close to replicating it. PayPal’s network remains strong, but its monetization potential is dwindling. Leveraging its network on unbranded processing is difficult by nature of the processing being unbranded and agnostic to whether you’re on the network or not.
PayPal has been trading alongside incumbent processors like FIS and Fiserv in the 7-11x TEV/EBITDA range since July of this year. This may seem harsh for a tech-forward business with serious network effect potential, but it highlights the uncertainty around what exactly PayPal will do at this moment to revive TPV growth, compete in the downmarket, and extract margin from a product that is otherwise in structural decline.
So is online payment processing a race to the bottom in the US? It would be if these processors weren’t coming at the secondary problems from so many different angles. PayPal, for one, has some very compelling assets to work with. Its ability to leverage its network on top of its growing unbranded distribution will define the story.
There’s a wider range of payment options and cross-currency situations that arise in the non-US market. Adyen differentiates itself by handling this really way as a “unified platform”. In the US, there are only a handful of commonly used payment methods, and multi-currency support is not a concern for the average SMB.
Paypal doesn’t break out branded vs unbranded numbers in their reporting. “Branded processing” includes the PayPal checkout button, Venmo, and several other platforms operating under different names
Q1F21 gave the first detailed segmentation TPV with Braintree. We can reliably back into Braintree volumes based on these numbers Credit: https://twitter.com/jkwade
Stratechery: An Interview with Lisa Ellis about payments: https://stratechery.com/2023/an-interview-with-lisa-ellis-about-payments/