Crossover Health might be the only health startup that can count Apple as its first customer. And for reasons we will get into below: Its most recent customer, Amazon, could be its last.

Crossover Health’s Origin Story

Picking a single event that led to the Crossover Health we know today is difficult because the company was woven from a number of threads that ran in parallel for years beforehand. I believe Crossover began to take shape the day that Dr. Martin C. Yee sent Dr. Scott Shreeve a direct message on Twitter in 2010.

But before we get to that, a little background: At the time, Shreeve and his two co-founders, Nate Murray and Dr. Rich Patragnoni, were running a direct primary care practice out of a single clinic located in Aliso Viejo, California. They wanted the clinic, called Crossover Health, to be less elitist than concierge care. So, instead of $2,000 per patient, they charged individuals $50-a-month and families $75-a-month.

The trio worked out pretty quickly that this model wouldn’t pan out, so both Shreeve and Patragnoni also got their insurance broker licenses. Their plan was to bolt-on some kind of catastrophic insurance to make the offering more comprehensive. As Patragnoni concluded in a reflective May 2020 interview:

A little over a month after the clinic opened its doors, Murray, who was the acting CFO back then, showed the doctors that it would take about 20 years for the business to grow to a sustainable number of patients. Luckily, the Crossover team soon got two lucky breaks.

First, they got a call from an executive at dialysis giant Davita who was interested in better understanding the direct primary care market. Davita hired Crossover for a consulting project that sought to better understand what a business model would look like for a scalable direct primary care practice. The company saw the consulting fees as a financial lifeline, but the project wasn’t quite enough to steer Crossover onto its eventual course: Clinics for self-insured employers. Here’s Patragnoni again:

Then Shreeve received that Twitter DM from Yee.

At the time, Shreeve had developed a following among people in the early Health 2.0 community thanks to his blog and Twitter account. So, it wasn’t unusual for Shreeve to hear from fans on Twitter. Yee was a physician with a practice in the Bay Area who spent one day each week at Apple’s campus in Cupertino doing preventive health exams. Yee told Shreeve in the Twitter DM that Apple was going to put out an RFP for an onsite clinic. He wanted to know if Shreeve and the team at Crossover wanted to take a crack at it with him.

A few months later Yee and the Crossover team were invited to formally respond to Apple’s RFP for the Apple Wellness Center. They were the clear underdogs. The competition was: Stanford, Palo Alto Medical Foundation, and CHS, which was the biggest on-site clinics company at that time.

Apple asked the team to respond to some 200 questions that included requests for 10 years of financials and 10 customer references. Crossover only had a few months of financials to share and zero references.

Shreeve told me in a recent interview:

In its original pitch, Crossover wanted to set up the Apple clinic using a PEPM model, which would essentially create a micro-capitation business model for Crossover. Instead, Apple and their outside consultants on the project insisted on a Cost+ model, which saw Apple paying for all the costs of setting up and staffing the clinic plus a management fee on top of that to Crossover. Patragnoni explained:

Soon after landing the Apple deal, Crossover got a call from Facebook, which had just finished up its own RFP process for an onsite clinic. Facebook said they didn’t like any of the worksite vendors that bid on it. They weren’t a “cultural fit”. And Facebook heard that Apple liked Crossover, so they invited them to pitch for their RFP too.

Crossover won that deal and, over the next couple of years, the company landed similar deals with Microsoft, Comcast, and other big tech companies.

Crossover’s beta customers: World-class engineers

As noted above, Crossover’s earliest customers gave the company a unique start. While landing Apple led to Facebook (and the combination of those two opened the door to many more big tech companies), Crossover also designed and iterated its earliest clinics with a lot of input from the employees of the two companies.

We will get into this more in the business model section further down, but because of Crossover Health’s business model, its practices didn’t need a traditional EHR. Crossover ended up building much of their software in-house. Shreeve told me:

Apple, of course, wanted the patient check-in process to start with an iPad. Facebook’s employees also had a lot of suggestions for how the software should work. Here’s Shreeve:

It wasn’t just the software that Apple and Facebook helped Crossover design in the early days. Shreeve was quick to admit his team didn’t have any experience building physical clinics. So Apple, true to its reputation, went deep on the details with Crossover:

Crossover Health’s Four Pivots

Crossover’s initial set of services were also heavily influenced by its first two customers: Apple and Facebook. Here’s Shreeve:

Worksite clinics have trended at various times throughout recent history, but for much of their existence they have largely offered occupational health-related services. (Most people summing up the legacy model for worksite clinics mention something about patching workers up after a worksite injury so they can get back to work.) By the time Crossover entered the market in 2010, worksite health clinics had already begun to look more like urgent care practices. Crossover was among the first to lean into both tech-enabled services and a more holistic approach to care that included primary care but also physical medicine, behavioral health, health coaching, and fitness programs.

Shreeve again:

(For some of its clients, Crossover also offers optometry and acupuncture.)

Over the past three years, Crossover also added care navigation capabilities as some of its customers asked it to help them manage the total cost of care for a portion of their employees.

While I think this service evolution is a helpful way to think about Crossover’s evolution since it got started with Apple eleven years ago, Crossover likes to describe its changes over the years as its “four pivots”.

2010. The first pivot was from DTC to self-insureds: This shift is covered at length in the origin story, but it’s when the Crossover team changed its model from a direct primary care practice that attempted to sell to individuals on the street to a worksite clinics company. (One of the keys to this model is that the self-insured employer pays for the clinic build-out and it typically operates under a Cost+ model, as mentioned in the origin story above.)

2015. The second pivot added near-site clinics: While Crossover’s first pivot truly was one, as the term is commonly used, its second pivot was a big move by the company, but it didn’t change its core self-insured employer-focused business model. Near-site clinics were also intended for self-insured employers as well as smaller ones that didn’t have a need for a dedicated on-site clinic all to themselves.

In October 2015, Crossover announced that it would launch its first “near-site” clinic in partnership with employee benefits consulting firm Towers Watson in Mountain View, California. From the announcement:

The Mountain View location launched with five employers already signed up, including Microsoft and Intuit. Despite the Towers Watson partnership at launch, I couldn’t find any mention of Towers in future Crossover near-site clinics’ launches.

Here’s how Shreeve explained near-site clinics to me:

2019. The third pivot was a move to virtual first: Soon after Crossover Health acquired virtual primary care provider Sherpaa, the company announced its third pivot, which would see it transition into a virtual first care provider.

Here’s how Shreeve distinguishes Crossover’s virtual-first strategy from others out there:

While Crossover’s patient-facing software has always been built in-house, the company’s system is built on top of Elation’s EHR. Crossover also layers in Health Catalyst’s full suite of data analytics. However, after Crossover acquired his company, Sherpaa’s founder and CEO Dr. Jay Parkinson spent two years leading the team that rebuilt Crossover’s patient-facing tech stack. Here’s how Parkinson described Crossover’s path forward for virtual first care in a recent podcast interview:

The implication from Parkinson is that over time Crossover’s physical clinics may shrink in size. New clinics are likely to be smaller than existing ones, and existing ones may even get a little smaller as time goes on and members start opting to go digital-first.

2021. The fourth pivot is Crossover becoming a lifestyle brand: In February of this year, Crossover Health’s CEO Dr. Scott Shreeve penned a blog post that announced the company’s fourth pivot was underway and it would become an aspirational lifestyle brand like Nike, Peloton, or Lululemon. While Crossover recently expanded its offerings into fitness classes, I’m not sure I really understand what this fourth pivot is all about. And if I do fully grasp it, I’m not sure this fourth pivot is as dramatic as the company’s past three pivots. Here’s how Shreeve explained it:

Crossover Health’s Growth Metrics

Like most private companies, Crossover Health generally keeps its annual revenue figures quiet. During his presentation at the all-virtual 39th Annual JP Morgan Health Care Conference, Crossover founder and CEO Dr. Scott Shreeve included a slide that disclosed the company’s annual revenues from 2017 through an estimated figure for 2021.

The chart below includes those numbers along with two I figured out for 2015 and 2016. One more historic revenue figure I didn’t include in the chart: Way back in 2012, Crossover brought in just $3 million in revenue.

The chart below tracks Crossover Health’s core business of standing up and running on-site clinics for self-insured employers as well as near-site clinics for satellite offices of big employers and groups of smaller local employers.

The total number of clinics for each year includes ones that were operational or close to launching (opening dates were hard to pin down for some of these). If a clinic shut down or changed hands in a given year, I counted them for the year of transition but not for the following year.

Take, for example, the years 2017 and 2018. On the face of it, Crossover Health had a flat year with no new clinic launches, but this was actually the period when Apple took over most of its on-site clinics and began to run the practices themselves via a wholly-owned provider practice called AC Wellness.

Apple wasn’t the only customer to go that route during these two years: H-E-B had set up two clinics through Crossover Health — under the brand Magenta Health — for its employees and other beneficiaries. The grocery store giant ended up cutting its ties with Crossover around the same time that Apple did. H-E-B went on to open other Magenta Health clinics in other parts of Texas after the separation.

So, while Crossover lost six clinics between 2017 and 2018, it managed to open six new ones too for a net gain of zero.

This chart also shows the impact of Crossover Health’s deal with Amazon. An incredible 17 of the clinics that Crossover Health filed paperwork to open in 2020 were for exclusive near-site clinics for Amazon employees and their dependents.

As far as I can tell, there are only 45 clinics that are open or in the works right now. As indicated below, Crossover said at the end of 2020 that it was operating 18 on-site clinics, 9 near-site clinics, and 4 virtual care-only clinics for a total of 31 at year-end.

Crossover Health’s four virtual care hubs around the country are in California, Texas, Florida, and (upstate) New York.

Crossover Health List of Customers

For on-site clinics, Crossover’s customers have included:

• Apple (no longer an on-site customer in California)
• Facebook
• Microsoft and Linkedin
• Comcast-NBCUniversal and Telemundo
• H-E-B (no longer an on-site customer)
• Visa
• Applied Materials
• Cummins
• Frost Bank
• Western Digital
• HP
• Micron
• Amazon

Crossover’s near-site and virtual-first customers include some of the employers above as well as:

• Intuit
• Square
• Veritas
• Synopsys
• NVIDIA
• Broadcom
• NFP
• Cisco
• Palo Alto Networks
• Pure Storage
• NetApp
• MUFG
• Dimensional

Revenue per clinic (RPC)

Crossover started out building and running on-site clinics for big tech employers before adding near-site clinics in late 2015. For back-of-the-envelope calculations, however, the revenue per clinic (RPC) for Crossover has always hovered around $3 million. Some years it is close to $2.7 million and others it gets up over $3.4 million, but it says right in that range. I figured this was a handy number to keep in mind as the company continues to grow, however, its recent move to transition into a virtual-first provider will likely make that RPC figure increasingly inaccurate. (More on Crossover’s four pivots in the business model section below.)

There’s a lot of assumptions baked into this, but I believe Crossover’s revenue per near-site clinic was about $3.6 million in 2020, but that may be including some small amount of virtual care-only revenue.

For its revenue per on-site clinic in 2020, I get about $4.4 million per clinic. Crossover had many Cost+ clinics in the works thanks to its big partnership with Amazon, but only 18 were operational at the end of 2020. Assuming its 2020 revenue did not include any revenue from Amazon (that’s a big assumption), then the $4.4 million in revenue per on-site clinic is about right.

Estimated PEPM pricing for 2020

Crossover Health posted $111 million annual revenue for 2020. About 29 percent of that came from its per employee per month (PEPM) agreements with employers for near-site clinic care and virtual care. That comes out to about $32 million. Assuming the PEPM is charged not by the member but by the eligible population, which is the figure Crossover share publicly, then we can get to a rough average PEPM for this part of Crossover’s business:

($32 million / 82,000 eligible lives) / 12 months = $33 PEPM

Crossover Health’s Business Models

Business model #1: DTC primary care clinic. As mentioned above, the initial business model for Crossover’s first standalone clinic was a direct primary care model that charged people $50-a-month and families $75-a-month. The Crossover team realized after four to six weeks of operation that it might take two decades to build the practice into a sustainable business this way. So, they began thinking about self-insured employers as a way to gain access to already aggregated populations of potential patients.

Business model #2: Per population Cost+ or fixed fee for on-site clinics. While Crossover tried to convince Apple to pay it a PEPM or PMPM fee for running its on-site clinic, Apple and its healthcare consultants were not comfortable with that model. Apple also wanted to invest quite a bit of money into the buildout of its onsite clinic, so it ended up paying Crossover for the costs of setting up the clinic and the costs of running it plus a management fee on top of that, which was Crossover’s margin. Some of Crossover’s on-site customers have moved into risk-based and total cost of care arrangements with them in recent years.

As of the end of 2020, some 71 percent of Crossover’s 2020 revenue came from its Cost+ and fixed fee agreements with its on-site clinic customers.

Business model #3: PEPM/PMPM. Once Crossover moved into near-site clinics and — eventually virtual-first primary care — the PEPM/PMPM became the company’s main business model. As of the end of 2020, Crossover said 29 percent of its annual revenue came from PEPM/PMPM agreements with near-site and virtual-first customers.

In a recent interview, Dr. Jay Parkinson indicated that both the Cost+ and the PEPM models required employers to shift a potentially significant slice of their healthcare spending from their medical budget to their HR budget. Since Crossover hasn’t traditionally worked with health insurance companies, its billing practices are typically accounted for as an HR expense by employers:

Here’s a slide from a presentation that Crossover Health’s CEO Scott Shreeve at the HLTH conference in 2019. While he didn’t explain it thoroughly during the talk, the slide represents the relative costs of three groups of patients. I’ll list them off right-to-left: those who are highly engaged with Crossover Health, those who use Crossover Health’s services sometimes, and a comparative group in the community who don’t use Crossover at all.

For the highly engaged group, the PEPM/PMPM numbers show that the costs of Crossover Health’s services for this population average about $50 PMPM, while the group’s total cost of care is around $250 PMPM. Those in the somewhat engaged with Crossover group use $100 PMPM in Crossover services but cost their employer closer to $375 a month.

While the implied savings was the point of Crossover’s slide, for the purposes of this report, it is an interesting hint at what Crossover charges its customers.

Business model #4: Risk-based for total cost of care. While I didn’t get many hard numbers for this one, Shreeve told me in a recent interview that Crossover typically is able to save employers about 15 percent for the cost of caring for those employees that fully engage with Crossover.

Starting sometime around January 2020, Crossover’s customers began inking total cost of care, risk-based contracts with the primary care provider. By then, Crossover had embraced its role as a care navigator, and this move to risk-based cost of care contracts really built on its standard PEPM/PMPM model. Around that time, Shreeve wrote a manifesto of sorts that articulated the company’s thinking around its business model:

In my recent interview with Shreeve, he pointed to the success of Medicare Advantage plans and suggested employers want their own version of that. They want a “Commercial Advantage” plan. That’s what Crossover is looking to provide with its risk-based contracts.

Finally, Crossover is set to launch a pilot program this summer (2021) with a still-unnamed health plan (Shreeve told me it is one of the big ones) that will allow it to test its Commercial Advantage plan concept with fully-insured employers.

Business model #5: DTC virtual first. This one is pretty straightforward. Crossover is once again piloting a direct-to-consumer primary care practice, but now it is virtual-first. Here’s what the pricing looks like in 2021 (a bit higher than the original $50-a-month from 2010). The company also noted that the listed services are just examples, not a comprehensive list:

Switching Costs and Crossover’s Lost Clinics

Before I did any digging into Crossover Health, I assumed that an on-site clinic was a business that had significantly higher switching costs than say, a virtual visits provider. Going into this report I was aware that Apple, Crossover’s first customer, had broken up with Crossover and taken over its on-site clinics to run the practices themselves.

I didn’t realize, however, (because it has never been reported) that Apple wasn’t the only Crossover customer defection.

The other big defection was one of Crossover’s Texas-based customers, H-E-B, a supermarket chain that operates 340 stores throughout the state of Texas and in Mexico and brings in about $25 billion in annual revenue. Crossover created two Texas clinics for H-E-B under the brand Magenta Health. H-E-B ended up breaking up with Crossover in 2018 and went on to open at least two more Magenta Health-branded clinics elsewhere in Texas.

Other clinics that Crossover has apparently lost or shut down include one in Indiana for diesel and alternative power company Cummins and one in Newburyport, Massachusetts for semiconductor services company Applied Materials.

Part of what surprises me about some of these shutdowns or clinic takeovers is that the provider teams often stay with the building and join the new practice under the new ownership. When Apple took over its California clinics, Dr. Martin Yee, who helped Crossover win the initial RFP to build clinics for Apple, ended up leaving Crossover Health in December 2017 to become a full-time employee at Apple as its clinics’ medical director in February 2018.

Crossover has benefited from this kind of transition too. In October 2017, Crossover won a bid to run Microsoft’s on-site clinic in Redmond, Washington, the flyer that Microsoft sent out to its employees in the area made clear that they would still be able to see their current doctor (as well as a long list of other members of the care team):

Crossover took over that clinic from one of the big names in work-site clinics, Premise Health, which has been in existence for 50 years. This phenomenon shows that the care teams at on-site clinics may technically be a part of Premise Health or Crossover Health, but they probably identify more readily with the employer and the employee population in their care.

It also indicates how easily a big company like Microsoft can switch out one on-site clinic provider for another: Microsoft’s clinic only shut down for nine days to accommodate the transition!

Crossover as Primary Care’s Re-Bundler and Navigator

Crossover Health believes it is well-positioned to win what it calls the ecosystem battle in employer-facing digital health. Here’s Shreeve:

Shreeve pointed to the many big digital health companies that started as a second opinions service or a cost transparency tool and moved into other services in a bid to become the center of the virtual health benefits ecosystem:

Shreeve also acknowledged that the COVID–19 pandemic accelerated employer adoption of digital offerings, but he said he thinks it will also accelerate them through the point solutions and to more comprehensive offerings like Crossover’s. As we’ll see in the next section, I’m not sure there is evidence of that yet, but it is a provocative prediction.

Amazon Deal vs Crossover IPO

Just as the founding of Crossover Health is inextricably tied to Apple’s health ambitions, Crossover’s exit strategy is bound up by its newest customer’s own healthcare plans: Amazon.

While 2019 and 2020 saw significant jumps in Crossover Health’s revenue, the company has not added any new on-site clinic customers since announcing its deal with Amazon in July 2020. The Amazon deal started out as a pilot but has grown to include plans for 17 clinics — operating under the Cost+ or fixed fee model — that will be located near Amazon facilities but exclusively for Amazon employees and their families.

Considering Crossover’s existing 18 Cost+ clinics generated 71 percent of its 2020 annual revenue — or $79 million — this deal with Amazon for 17 similar clinics (assuming it won’t grow beyond that) is likely to generate $74 million in annual revenue.

That leads to a scary question for a company that is on a shortlist for digital health IPOs in 2021: Will Amazon soon make up close to half of Crossover’s annual revenues?

Amazon’s own healthcare ambitions under the brand Amazon Care have been well-documented in recent months by STAT. In 2021, the generically-named medical practice that powers Amazon Care, Care Medical PC, filed paperwork to set up business entities in all 50 states and in Washington D.C.

Meanwhile, Crossover Health has set up its four virtual care practices to cover the country with its own nationwide primary care practice. Much of that coverage came as a result of Crossover’s acquisition of Sherpaa.

However, Crossover has also been setting up business entities in various states. By my count, Crossover has set up shop (on paper, at least) in 29 states, and at least 18 of those new business entities were set up in either 2020 or 2021.

For example, Crossover Health just formed a medical practice subsidiary in Alaska on April 12, 2021.

Clearly, Crossover Health and Amazon Care are two partners that are on a collision course of some kind, right? Either Amazon acquires Crossover and launches Amazon Care on the book of business that Crossover has carefully built up over the past decade or Amazon eventually pulls out of its clinics partnership with Crossover — just like Apple and H-E-B did.

Is there a scenario where these two companies remain partners? And if they did, and Crossover Health were to go public with its current book of business, how would Wall Street view its partnership with Amazon? It contribution to Crossover’s revenue is already huge, but it might get much bigger:

This forward-looking slide from a recent Crossover Health deck notes that its deal with Amazon could expand well beyond the five geographic regions it is currently building 17 clinics into another 15 regions. If the math holds (3.4 clinics per region), Crossover could be building another 51 clinics for Amazon across these 15 new regions to blanket the country in Crossover-run clinics for all of Amazon’s US-based employees and their families.

Using the $4.4 million estimate revenue per Cost+ clinic, that might be another $224 million in revenue from Amazon once this deal is fully executed! It’s not clear, however, that Crossover has won this expanded deal yet.

Beyond Amazon, the slide above points to Crossover’s other future plans: The company clearly sees its virtual first strategy and its upcoming pilot (summer 2021) for a Commercial Advantage plan offered through a big health insurance player as important levers.

E&O will be tracking those two initiatives closely while keeping an eye out for an expanded deal announcement with Amazon and a Crossover Health S-1 filing.