Staff - Ai: Pricing For Disruptive Technology

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

STAFF.

AI: PRICING FOR DISRUPTIVE TECHNOLOGY

On a brisk fall day in 2018, Dr. Carl Einarsson sat at Boston Logan International Airport waiting
to board a flight to Dallas, Texas to meet with senior management of Pillar Healthcare, a for-
profit hospital system. While waiting, he reflected on his recent meeting with State General
Hospital (SGH). “The CFO loves the financial appeal of Staff.AI and the head of Human
Resources loves the time savings possible,” recalled Einarsson. “But the best reaction was from
the unit coordinators and schedulers—Staff.AI is going to make their job a lot easier.”

Staff.AI, the brainchild of Einarsson, was a resource-sharing application that matched physicians
with hospital staffing requirements. Operating similar to Airbnb, Uber, and other matching
applications, Staff.AI featured web-based desktop software allowing human resource
administrators to post hospital shift requirements and a parallel mobile app for physicians to
review and sign up for posted shifts. Staff.AI significantly decreased the time and cost
associated with temporary staffing agencies performing the same task, while offering much
greater convenience to both physicians and hospital administrators.

The timing for Staff.AI was favorable, as other resource-sharing applications had experienced
substantial success in the marketplace. Staff.AI presented significantly greater opportunity in
health care, however, as clinical quality carried higher value for patients and health plan sponsors
than other applications. Moreover, the target for Staff.AI’s services was the entire health care
payroll, projected to be $728 billion in 2018.1

As a surgeon, however, Einarsson knew there was a deeper, more important value of Staff.AI:
matching operating room teams that worked together seamlessly, delivering the best patient
outcomes. Staff.AI could literally save lives. Although the financial prospects of Staff.AI were
appealing, Einarsson wanted to change health care.

In light of this, Einarsson pondered what the best revenue model for Staff.AI might be. “Should
we aim for rapid market introduction, continuing the beta-test approach to build scale, or should
we establish a pricing model for entry into the for-profit hospital environment?” thought
Einarsson. The first path would require substantial investor capital, trading sales revenue for
early growth, while the second would deliver accelerated earnings, important for exit value. “If
we pursue the latter, is the best model a single-sale product approach, a service-based
subscription fee, or something different?” These questions weighed on Einarsson’s mind as he
prepared to board the plane for Dallas to meet with the senior team at Pillar Healthcare to discuss
how Staff.AI could improve scheduling of physicians and save the hospital money. At the end of
the day, he knew he they would want to know the price. What was he going to tell them?

Background
Staff.AI was an innovation at the intersection of health care and software, industries that
witnessed substantial change in the two decades since the late 1990s. Although both experienced
strong technological trends, each was impacted differently: where software drove improved
productivity and decreased costs, a direct consequence of Moore’s Law (the principle that
computing capacity doubles about every two years), the explosion of innovation in medical
devices and pharmaceutical products had the opposite effect in health care, spiraling costs at a
multiple of ordinary inflation. Moreover, intense regulation had stalled productivity and quality
improvements in health care.

Resource-sharing and mobile technology


A significant beneficiary of software success was telecommunications. Displacing everything
from news reporting to retail sales to running errands, the combination of the Internet and mobile
technology changed how consumers and service providers interacted on a regular basis. One
particularly useful application was the category of resource-sharing activities.

As transaction brokers, resource-sharing companies benefited from relatively low capital


requirements and increasing returns from adoption, leading to auction pricing models and low
cost of trial. Airbnb disrupted the hotel industry and TaskRabbit did the same thing with errands.
When ridesharing company Uber introduced its UberX app to drivers in July 2012, resource-
sharing took off. By 2018, over 134 million people used resource-sharing applications
nationwide.2,3

Einarsson was intrigued by the success of the resource-sharing models, because of both their
operating and financial efficiency. “The most common element of all of these disruptors was the
matching of parties; those offering and needing a service are able to obtain more precisely what
they want more efficiently,” he reflected in the spring of 2017. “Improving outcomes by
matching parties was brilliant. Would the same thing work for highly skilled professional
services like medical care?”

Hospitals and health care


Hospitals and health care technology charted a different path. With the introduction of Medicare
in 1966, U.S. health care began a divergence from outcomes and results to price and cost control.
From 1967 to 2018, annual growth in the cost of health care averaged 5.6%, compared to just
3.1% per year from 1946 to 1966.4 In addition, introduction of legislation such as the Employee
Retirement Income Security Act of 1974, the Tax Equity and Fiscal Responsibility Act in 1982,
successive Budget Reconciliation Acts from 1985 through 1989, and the Balanced Budget Act of
1997 all combined to decouple health care consumers and providers, such as physicians and
hospitals. The net effect was escalation of demand for health care and an accompanying
decrease in access and quality.

At the same time, introduction of new medical products skyrocketed, with 670 new
pharmaceutical products and 2,192 new medical devices approved by the U.S. Food & Drug
Administration from 1997 to 2018.5,6 The combined increase in regulation, distancing of patient
and provider, and explosion in new technologies resulted in a perfect storm of rising costs and
diminishing outcomes, eventually leading to passage of the Patient Protection and Affordable
Care Act in 2010, commonly referred to as Obamacare. The result was a significant decline in
health insurance coverage and rise in deductibles, leading to an increase in the underinsured,
climbing to 28% of all individuals in 2016.7 Worse, amenable mortality (deaths otherwise
avoidable with proper care) in conditions such as hypertension, kidney and heart disease,
diabetes, and stroke increased.8

2
All these factors had the effect of decreasing hospital and physician productivity, and placed
stress on hiring, scheduling, and managing medical professionals throughout the health care
system. Medical payroll, at 20% of total health care costs, was an escalating issue.

Hospital Staffing
While technology was changing telecommunications and health care, hospital staffing was
slower to advance. Like the medieval European monasteries of St. Giles, St. Anthony, and St.
Leonard that tended to the sick, modern facilities were still staffed with caregivers tending to the
spiritual, hospitality, and research needs of the community. The term hospital evolved naturally
from the hospitality mission of these early establishments.

Traditionally associated with religious institutions, diagnostic, operating, and other specialized
activities were added as technology advanced. Large community hospitals developed into
centers of research like Vanderbilt Medical Center in Nashville, for example, while for-profit
hospitals arrived late on the scene in the 20th century, morphing into large, managerially focused
operations such as the national chain – HCA Healthcare. This factored into Einarsson’s thinking
as he considered Staff.AI’s value proposition in late 2017:

“The evolution of care has led to very complex hospitals with differing goals. For instance, an
academic medical center, typically large in terms of revenue, treats the most complicated cases
with a deep bench of physicians to perform the care. Individual community hospitals, or a health
system comprised of a number of smaller hospitals, treat less complex patients with fewer
resources, but are still required to maintain a comprehensive panel of specialists. Lastly, not-for-
profit health systems also must attract and maintain adequate physicians with admitting
privileges, but also exhibit the best profit margin. All of these factors impact hospitals’ ability to
invest in new technology, while struggling with complex business organizations dwarfing the
size of the clinical staff.”

(See Exhibit 1 for a typical hospital organization chart and Exhibit 2 for an illustration of the
financial performance of different types of hospitals.)

Through the evolution, hospital staffing remained the same. To deal with shifting demand, the
major innovation was the temporary, or locum tenens, physician. This category of health care
professional was an individual brought on board to assist during periods of higher than normal
activity. Other positions eventually followed, and temporary unit coordinators, nurses, and other
non-medical personnel were staffed to match demand. A typical hospital relied on temporary
personnel for 1%-3% of staffing needs across all categories of activity (i.e., total payroll).9,10

Temporary staffing solutions


To meet the varying demand for staffing, hospitals utilized on-call personnel who could come to
the facility when needed. If a requirement was of longer duration or local supply was limited,
locum tenens physicians and temporary personnel were often utilized. These individuals were
typically identified and placed by specialized temporary staffing firms, companies such as Cross
Country Healthcare, focusing on the unique requirements of each hospital. Contract periods
ranged from one to six months with compensation driven by available supply; the staffing
agency typically charged 20%-100% on top of ordinary compensation.11 This was an important
factor for hospitals with varying acuities—an academic medical center, for example, would

3
spend much more on highly compensated specialist physicians than a small community hospital
providing less complex care. Initial credentialing, if needed, was usually done by the agency,
while certification and training was the responsibility of the hospital.

A number of traditional health staffing firms had developed applications to automate matching of
jobs and individuals, but only one, ROVR, had moved beyond matching to offer scheduling
services. None had envisioned analytics to measure and match professionals for the best patient
outcomes, a significant differentiator of Staff.AI.

The move to automation predictably ate into traditional firm profits, as they raced to implement
new technology while weathering volume declines under increasingly competitive pricing. (See
Exhibit 3 for data on financial performance for leading temporary physician staffing company
Cross Country Healthcare.) In its 2018 Annual Report, Cross Country Healthcare reported a
trifecta of bad news, with revenue declining faster than both operating and selling, general, and
administrative expenses, while write-downs of overly rosy acquisitions and restructuring charges
added to the company’s woes.12 “It was clear that technology was impacting the cost of health
care staffing,” Einarsson mused, “but was it helping in other ways?”

Impact of staffing on health outcomes


Hospital staffing and scheduling decisions affected not only cost and financial performance, but
also clinical outcomes of care. In addition to improving communication during procedures,
effective medical teams across the treatment pyramid (physicians, nurses, medical aides, etc.)
had a large impact on adverse events, treatment efficacy, length of stay, and recovery periods. In
one study of care performance on acute trauma patients, researchers found a substantially greater
time to disposition in low-teamwork operating groups (35 minutes) compared to high-teamwork
groups (20 minutes)—sometimes the difference between life and death.13

Although all hospitals employed some type of scheduling system, the largest facilities utilized
software provided by two health information technology firms, Epic Systems and Cerner
Corporation, focused chiefly on electronic health records (EHRs). Neither provided more than
minimal resource scheduling functionality. Moreover, neither Epic nor Cerner provided
functionality to identify and recruit medical professionals. Developments in health information
systems had grown considerably, driven by meaningful use of electronic medical records
required in the American Recovery and Reinvestment Act of 2009; however, little innovation
occurred in the measurement of team effectiveness in the clinical environment. This represented
a substantial unmet need in the delivery of care, and a need Staff.AI’s scheduling capabilities
directly enabled.

Staff.AI
Einarsson, a graduate and former neurosurgical resident of the University of Cambridge,
completed his master’s degree at Harvard University in 2018. Between the programs, Einarsson
worked at consulting firm McKinsey & Company, where his entrepreneurial interest was
sparked. As the capstone project at Harvard, in the spring of 2017 Einarsson developed a
business plan for Staff.AI that would alleviate the cost and streamline scheduling of hospital
medical teams.

4
Einarsson knew the problems with scheduling in the hospital environment, and as a highly
trained surgeon in one of the most demanding specialties, neurosurgery, he was also aware of the
impact of poor staffing on care outcomes. With his master’s degree project, Einarsson leaped at
the opportunity to address this, committing the idea for a new staffing company and resource-
sharing application to paper. Einarsson sketched out the model for desktop and mobile interfaces
and hired local students to code them. By the summer of 2017, Einarsson had a working version
of the application.

A requirement of Einarsson’s project included a beta-test with a potential customer, and his
course advisor introduced him to the CEO of State General Hospital, Richard Masters. In the
initial meeting, Masters explained the operational and economic aspects of temporary physicians:

“We spend millions of dollars a year on temporary employees,” Masters said. “It’s expensive,
too—half of our costs, $2 billion, are payroll and 1% of that for temporary employees is $20
million—at a 20% staffing agency mark-up, that’s $4 million not being spent on care.” Masters
continued, “And that doesn’t even address the quality aspects of having the right team driving
the care.”

After the meeting, Masters arranged for Einarsson to meet with the SGH chief of human
resources who agreed to test the product with their internal staff. Working with an enthusiastic
advisor, Dr. Aarav Chowdhury, Einarsson trained the staff on how to load personnel and match
them with activities in the various areas of the hospital. Soon the staff was using Staff.AI to
schedule all professionals. (See Exhibits 4 and 5 for a diagram of the process flow and
screenshots of the various Staff.AI interfaces.)

The product was embraced, and Einarsson worked diligently to incorporate suggested
improvements to the software. As an integrated online resource comparing favorably both
operationally and financially to temporary staffing agencies, hospital management liked the
value of Staff.AI, and internal staff liked the flexibility of the real-time, online scheduling
platform. A surprise, however, was the value placed on the originally conceived matching
functionality. Reflecting back to the SGH meeting, Einarsson theorized,

“It was encouraging to see our first customer, SGH, rave about the streamlining of scheduling. It
was discouraging, however, to see their apprehension about the external recruiting we introduced
initially. Discussion and further development of matching a specific physician with a specific
team, providing analytics based on outcomes data, crystallized our product value. More
pointedly than either resource-sharing or traditional staffing services, it was this ability to match
resources for higher quality that seemed to be driving user appeal.”

Uncertain about how to approach payment for the product, Einarsson demurred, delaying
definitive conversations about pricing for SGH. By the end of the summer after graduation,
energized by the results at SGH, Einarsson made the decision to pursue the project full time,
launching the company to deploy it.

Operating and deployment costs


As with all new ventures, Staff.AI faced a challenge: how to manufacture and deliver the
product cost-effectively. Fortunately, the product’s value was considered positively by virtually
all stakeholders, especially given the rates charged by temporary staffing agencies. Maximum

5
value, however, would be gained by integrating with external hiring, and that might include
additional costs for recruiting, credentialing, advertising, and promoting skilled physicians. To
Einarsson, the challenges appeared to be (a) ramping up installed customers to establish first-
mover advantage and (b) recruiting the capital to achieve this.

Einarsson met with a number of venture capital (VC) firms and hospital technology consultants
to try to determine a general pricing approach. His research was instructive in its lack of
consensus:

“The managing director of a major health care VC firm indicated a typical hospital software
system would be priced as a subscription product in the six-figure [$100,000 per year] range,
while another technology VC firm pointed toward a percent-of-savings model; yet another firm
said to ‘offer the product for free—like Facebook—to build scale.’ At the same time, SGH was
hinting at its interest in an ownership stake in Staff.AI in our discussions. One thing was clear:
there was no easy way to ‘set the price.’”

Meanwhile, Einarsson had to fund operations during the beta-test. Making rough assumptions
about fielding cost and relative value to customers, Einarsson and his advisors developed a
subscription-based deployment model that they thought could build a substantial footprint in a
short period of time (see Exhibit 6). With a total addressable market of over 8,000 hospitals
nationwide, Einarsson set half a percent penetration, or 40 installations, as his target. Based on
these assumptions, Einarsson estimated this would require investment capital of at least $1
million.

In the fall of 2017, while the SGH beta-test continued, Einarsson took his business plan on the
road to investors. By June 2018, on the heels of the success at SGH, Einarsson secured an
invitation to apply for a grant from the European health ministry he had been working with, as
well as tenders for investment totaling $2 million. With short-term funding of the company
assured, Einarsson was ready to begin developing new clients.

Leveraging the success with SGH, Einarsson secured a meeting with Pillar Healthcare, a leading
publicly traded U.S. hospital firm, operating 89 acute care facilities nationwide. It was clear
from preliminary conversations that Staff.AI could provide substantial value for Pillar, but
Staff.AI’s cost model was still in flux: deployment was very sensitive to user circumstances,
things like existing technology investment, market-specific labor costs, and staffing agency
pricing.

The question remained, however—should Staff.AI continue to be offered using a beta-test


approach, giving his product away to build customers and gain access to valuable outcomes data,
or should Einarsson implement an early, reasoned, pricing strategy? “Staff.AI means different
things to different stakeholders, and hospitals derive widely varying value from it,” thought
Einarsson. “There really isn’t a roadmap for this.”

The Decision
Much had happened with Einarsson and Staff.AI over the past two years—Einarsson would
never have guessed he would make it this far in such a short period of time. And yet, the early
successes brought increasing challenges, and Einarsson knew the next decision to price the
product would significantly impact its trajectory going forward. His mind wandered with the

6
planes passing the great windows of the gate area at Boston Logan: would Staff.AI be better off
with a quick, accelerated takeoff, giving the product away for rapid adoption, or would its
success be maximized by a value-added pricing model and a longer, more deliberate ascent?

“Boarding will begin shortly for Jet Blue Flight 1115 to Dallas-Fort Worth at Gate C42,”
announced the attendant over the gathering crowd. Einarsson came back to the present, recalling
how he had gotten here with Staff.AI. His feeling of pride was interrupted by the cool blast from
the jetway, as the door leading to the aircraft opened.

“SGH has gone well,” thought Einarsson, “but their focus on internal staffing barely touches the
value of resource-sharing.” The enthusiasm of the SGH staff was certainly positive, and their
interest in outcomes-based team matching dovetailed with Einarsson’s goals of improving health
care, but the internal focus failed to monetize the value of disrupting temporary staffing agencies.

On the other hand, Pillar Healthcare was a profit-driven entity, answering to shareholders that
liked investment returns and growth. “At 1% of total costs, reducing temporary staffing would
flow right to the bottom line,” mused Einarsson.

Was another beta-test appropriate for Pillar, or was it time to introduce a pricing model? If
pricing was appropriate, was enough information available to set a reliable price attractive to all
customers? Checking Pillar’s financials on his iPhone, Einarsson wondered, “$18.2 billion in
total costs, 1% temporary staffing, and stock trading at 16 times earnings… that’s $2.9 billion of
shareholder value—a chunk of cash.”

“Come on, I need your boarding pass,” said the attendant. Einarsson began to settle on what he
would propose to Pillar as he stepped onto the jet bridge.

7
Exhibit 1: Sample Hospital Organization Chart

Source: Compiled from interviews with company.

8
Exhibit 2: Illustrative Financial Data by Type of Hospital

Note: Numbers may not add to 100% due to rounding.

Source: Compiled from interviews with company.

9
Exhibit 3: Financial Statements of Cross Country Healthcare

Source: Cross Country Healthcare, Inc. (2018). Annual report 2018. Retrieved from
http://ir.crosscountryhealthcare.com/financial-information.

10
Exhibit 3: Financial Statements of Cross Country Healthcare (cont.)

Source: Cross Country Healthcare, Inc. (2018). Annual report 2018. Retrieved from
http://ir.crosscountryhealthcare.com/financial-information.

11
Exhibit 4: Staff.AI Process Flow

Source: Company.

12
Exhibit 5: Screenshots of Staff.AI User Interfaces

Hospital Dashboard

Source: Company.

13
Exhibit 5: Screenshots of Staff.AI User Interfaces (cont.)

Hospital Scheduler

Source: Company.

14
Exhibit 5: Screenshots of Staff.AI User Interfaces (cont.)

Mobile App

Source: Company.

15
Exhibit 6: Staff.AI Initial Investment and Deployment Plan

Source: Company.

16
Endnotes

1
Centers for Medicare & Medicaid Services. (2019). NHE Fact Sheet. Retrieved from
https://www.cms.gov/research-statistics-data-and-systems/statistics-trends-and-
reports/nationalhealthexpenddata/nhe-fact-sheet.html.
2
Statista. (2018). Ridesharing services in the U.S. - Statistics & Facts. Retrieved from
https://www.statista.com/topics/4610/ridesharing-services-in-the-us.
3
U.S. Census Bureau, op. cit.
4
Bureau of Labor Statistics. (2019). Medical care in U.S. city average, all urban consumers, not
seasonally adjusted. Retrieved from https://data.bls.gov/cgi-bin/srgate.
5
U.S. Food & Drug Administration. (2019). New Molecular Entity (NME) Drug and New Biologic
Approvals. Retrieved from https://www.fda.gov/drugs/nda-and-bla-approvals/new-molecular-entity-nme-
drug-and-new-biologic-approvals.
6
U.S. Food & Drug Administration. (2019). PMA Approvals. Retrieved from
https://www.fda.gov/medical-devices/device-approvals-denials-and-clearances/pma-approvals.
7
The Commonwealth Fund. (2017). Underinsured Rate Increased Sharply in 2016; More Than Two of
Five Marketplace Enrollees and a Quarter of People with Employer Health Insurance Plans are Now
Underinsured. Retrieved from https://www.commonwealthfund.org/press-release/2017/underinsured-rate-
increased-sharply-2016-more-two-five-marketplace-enrollees-and.
8
The Lancet. (2017). Healthcare Access and Quality Index based on mortality from causes amenable to
personal health care. Retrieved from https://www.thelancet.com/journals/lancet/article/PIIS0140-
6736(17)30818-8.
9
Blumenthal, D.M., Tsugawa, Y., Olenski, A., & Jena, A.B. (2017). Substitute Doctors Are Becoming
More Common. What Do We Know About Their Quality of Care? Retrieved from
https://hbr.org/2017/12/substitute-doctors-are-becoming-more-common-what-do-we-know-about-their-
quality-of-care.
10
Staffing Industry Analysts. (2018). 2018 Benchmarking Survey Provides Insights Into US Allied Health
and Advanced Practice Staffing Markets. Retrieved from
https://www2.staffingindustry.com/eng/Editorial/Healthcare-Staffing-Report/Nov.-8-2018/2018-
Benchmarking-Survey-provides-insights-into-US-allied-health-and-advanced-practice-staffing-markets.
11
LocumTenens.com. (2020). Locum tenens cost comparison calculator. Retrieved from
https://www.locumtenens.com/resource-center/interactive-tools/cost-comparison-calculator.
12
Cross Country Healthcare, Inc. (2018). Annual report 2018. Retrieved from
http://ir.crosscountryhealthcare.com/financial-information.
13
Pucher, P.H., Aggarwal, R., Batrick, N., Jenkins, M., & Darzi, A. (2014). Nontechnical skills
performance and care processes in the management of the acute trauma patient. Surgery, 155(5), 902-909.
doi:10.1016/j.surg.2013.12.029.

17

You might also like