In this blog, I look at how NTT DATA worked with a large Blue Cross Blue Shield (BCBS) health insurance carrier to establish an enterprise governance structure for automation, and at the lessons learnt along the way.

Like many other large BCBS carriers, the company had piloted RPA initiatives, and from the somewhat frustrating results of these experiments, it had formed two conclusions:

• An IT department-driven center of excellence delivering bots will not achieve the full potential of automation
• Point solutions being driven within individual towers/business units are not scalable across the enterprise.

The company concluded that before it could proceed with its automation journey, it required an automation governance structure that aligned with the enterprise strategy. A business-driven (rather than IT-driven) deployment of RPA needed to coordinate the needs, requirements and deployment of RPA across the front, middle and back office functions, as well as shared and internal ancillary services.

The BCBS carrier hired a team from NTT DATA, led by Deana Rhoades, the Global Practice Lead, Healthcare Automation “to create an enterprise-wide governance structure customized to their corporate strategic objectives and organizational culture”. Within the context of the enterprise’s goals, strategy, and current workforce, the company tasked NTT DATA to create the automation strategy, the decision frameworks and the organizational structure. While the BCBS company had long before established solid objectives, frameworks and management systems for its human workforce, the company realized it needed to lay the foundation for the same kind of structure for automation (and the bots) of its “digital workforce”.

Starting in August of 2018, NTT DATA began its work creating an enterprise level governance structure for automation. It focused on scalability considerations and governance, treating bot development “almost as an afterthought”. The tactical view about how to purchase and deploy automation solutions and build bots on different platforms would flow from the enterprise’s strategic objectives and from appropriate integration of the human and the proposed digital workforces. It took two months for NTT DATA and its client to articulate the following governance model, composed of three layers:

Layer 1: Sponsorship

Champions of the RPA transformation articulated the vision and goals for the automation journey and monitored performance of the COE. Sponsors include high-level representatives of the COO, the CIO and the HR departments, coordinated by a Program Management Office (PMO). Strategic frameworks now articulate the enterprise’s objectives, categorize potential automation projects within that context, and facilitate decisions about deployment in terms of (for example):

• Potential cost savings (prioritized over revenue)
• User experience (prioritized over productivity).

Layer 2: Enterprise Capability Center

This team unites leaders and dedicated resources from the following functions: HR, Data and Analytics, IT, Security, Organizational Change Management, Business Process Management, and Operations. Six workgroups develop and provide expertise on the core COE capabilities. The COE subgroups cascade the automation strategy into action plans that provide capabilities across automation development teams and business units. Focus areas include:

• Strategy and Measurement – turns strategy into executable components; owns success criteria, key performance indicators (KPIs) and objectives and key results (OKRs); quantifies the value of the COE
• Pipeline Management – generates demand for automation at the process level among BCBS company employees, prioritizes and schedules the resulting workstreams
• Workforce Strategy – defines needed FTE skills and gaps, owns the organizational change management (OCM) plans and provides training for BCBS company employees
• Automation Standards – develops the standards, tools, repositories, policies and procedures that guide all automation initiatives
• Data Strategy – maintains data management strategy, defines how automation software accesses and collects data, and how the automation efforts comply with risk and security policies
• Virtual Workforce Monitoring – maintains a centralized command center to monitor and oversee the bots in production.

Layer 3: Automation Factory

Delivery and deployment teams work under the aegis of the leadership priorities and plans developed in layers 1 and 2 with complementary aims:

• Demand generation – generating awareness and demand for automation within the enterprise at the level of the teams that manage specific processes. A change management team trains these teams on capabilities of RPA and helps them see the value of implementing the technology

• Technology delivery – agile development teams automate processes using the appropriate tools and platforms, such as Blue Prism and UI Path.

For the next phase of work, NTT DATA has begun to create a complementary hybrid (or “federated”) operating model for agile delivery of bots. This hybrid model is supposed to establish the guardrails and frameworks needed by individual business units that have the skills and the desire to build their own bots. The hybrid model is expected to augment the centralized enterprise governance model by 2020.

The human response?

With NTT DATA, the BCBS company has worked to communicate with various business units and with their leaders to resolve their questions and any potential anxiety about the use of bots. During the BCBS company’s prior work with another IT consulting firm, it had developed its own home-grown automation tools. The in-house deployment of an RPA platform had introduced the company to concepts and practices at a tactical level. Activities surrounding these pilots had been widely broadcast through various communication channels, including robotic roadshows, Yammer, and email.  As a result of this in-house publicity, NTT DATA reported that it met with more curiosity and less resistance than expected. NTT DATA also reported that company business units and employees had already begun to form opinions about automation through the lens of their experience with their prior RPA tool, opinions that needed to be considered if and when other development tools were introduced.

The business consequence?

NTT DATA believes that the BCBS carrier has taken a significant stride up the automation maturity curve by articulating a governance model with the following elements

• Charter
• Roles and responsibilities
• Leaders
• Change management
• Resources dedicated to organizational communication and demand generation
• Resources dedicated to development of a broader set of intelligent automation technologies.

RPA initiatives that predate the NTT DATA-led exercise in defining automation governance now have a structure and resources available when they need to escalate issues, and have realized greater ROI. Furthermore, the BCBS carrier’s “ox in the ditch” initiatives have now been organized into six workstreams, and in future the company believes that its governance structure and measured approach will yield expected ROI and that its human and virtual workforces will complement each other efficiently.

In my previous blog, I described how, since 2017, Sutherland has created a shared services model that obviates the need for participating California health plans to separately build and update parallel databases to track the availability of providers of nonurgent care for Medicaid recipients.

The company estimates that through its consortium of member health plans it has reduced associated health plan physician data management costs by 75% through elimination of duplicative work and by improvement in survey execution workflow and other areas. For an estimated 80,000 physicians in its CA directory, Sutherland now estimates that it reduces the touch rate on providers related to the Provider Appointment Availability Survey (PAAS) from three to one call per practice. The initiative also improves reporting and other interactions with the California regulatory body (Department of Managed Healthcare, or DMHC) and improves patient access to timely care.

Sutherland’s success with its coopetition/shared services model begs an interesting question: can this model be extended across the U.S. and, if so, how?

Uncovering value from duplicated effort

The coopetition model now proven in California might provide a useful template for future work at the national level. Data from Sutherland’s efforts in California indicate that national health plan provider networks significantly overlap and that much of the work they pursue in building and maintaining their physician databases is therefore duplicative and wasteful. In California, Sutherland reports a 48% overlap of providers between the top three CA health plans. That is, of ~20,000 physicians that are currently contracted to plans managed by one of the top three health plans in CA’s Medi-Cal Medicaid program, over 9,000 are currently contracted with all three health plans. Each health plan in California is required by the DMHC to maintain accurate data on each provider so that patients can gain access to timely care. Each health plan is further required to manage this dataset in order to maintain its own operations. The difficulties in maintaining these parallel datasets result in a myriad of problems for different stakeholders, including wasted effort.

Stakeholders include vendors of business outsourcing services. Prior to Sutherland’s involvement in the shared services initiative, the data collected by the DMHC was of such poor quality that it resulted in a directive to all CA health plans saying that the vendor then in charge of managing the provider data collection effort would no longer be allowed to work in CA.

Sutherland reported that, at that time, 40% of data records contained errors or omissions. The result was that health plans could not confirm members for timely and appropriate access to care, and providers were subjected to unnecessary inconvenience, cost and fatigue. The opportunity for a vendor of business outsourcing services, conversely, was significant. Since two-thirds of data collection efforts by different health plans required the same basic information from providers, Sutherland identified an opportunity in California to generate value by eliminating unnecessary work and collecting a slice of the resulting value, while simultaneously providing value to the regulatory body, providers, and patients.

Geographic & market segment extension of the model

The geographic extension of this model in physician network data management beyond California may be a logical next step. Sutherland itself calls its shared services model for the provider appointment availability survey (PAAS) a “proof of concept”. The fact that Sutherland has successfully united the interests of competing health plans with those of providers, patients, and the state regulatory body lends credence to the idea that other health plans in the U.S. might be convinced to join a similar consortium. Note that some health plans would likely never be candidates, such as Kaiser Permanente, which is based on a vertically integrated model that unifies the management of provision and reimbursement of care. (While Kaiser provides Medicaid services in California, it is not a member of Sutherland’s current shared services model in CA).

However, whether led by Sutherland or another entity (private or public sector), such a consortium could eliminate waste on a state-by-state basis, or even more broadly. The model could be extended to other government healthcare. It could standardize and streamline data collection, present accurate data to a wide range of stakeholders in timely fashion, standardize reporting, reduce provider fatigue significantly, and improve customer/patient access.

Generating leverage

Creating a public utility by mandate may lead to inefficient, unintended consequences, but Sutherland’s success seems to indicate that a market solution can be viable. The CA consortium currently counts 14 health plans, but replicating this success outside CA would require customization to other economic and political circumstances. The mission of the Council for Affordable Quality Healthcare (CAQH) and other associated alliances, non-profits, and government agencies may align with such efforts. Companies that specialize in providing outsourcing services have, as Sutherland proves, many of the capabilities required. Short of a government-sponsored mandate, how can health plans be induced to share proprietary data and data methodologies?

Political leverage might be hard to generate among consumers/patients, but physicians may present a more unified and sharply-focused interest group. If a doctor contracts with a single health plan for multiple products (e.g. Medicare Advantage, Mental Health, etc.) and that doctor’s information needs to be verified for each product, this would require multiple touches, cost, inconvenience, and fatigue. According to Sutherland’s experience in CA, that doctor may, on average, contract with 20 health plan products. The doctor is therefore incentivized to reduce this duplicative and wasteful interaction, and the argument that physician rosters can be harmonized among health plans with minimal interaction (leveraging web portals rather than call centers) is not hard to make. Having thus grasped the challenge, the physicians’ professional organizations may be well-placed to work with health plans to set up more consortia similar to Sutherland’s in California.

Finding allies

An industry alliance designed to introduce blockchain is aimed directly at the challenge of reducing the estimated $2.1 bn in cost associated with maintaining provider data. According to an April 2018 healthcareITnews.com article, Optum, UnitedHealthcare, Humana, others launch blockchain pilot, these industry titans are exploiting the opportunity to reduce waste associated with provider data: “Five healthcare organizations including insurers UnitedHealthcare and Humana, Optum, Quest Diagnostics and MultiPlan are launching a blockchain pilot to help payers tackle mandated provider directories”.

The mission of this alliance may provide a long-term objective to which one or more consortia based on the Sutherland CA model might be mutually supportive. The hype associated with blockchain might create the attention necessary to establish more provider data consortia, while the political clout of physicians’ professional organizations might bring leverage. In combination, private sector players might then find the resources and support necessary to align economic incentives, manage workflows, normalize and de-duplicate data, execute against state and federal regulations, and package provider data in digestible, accurate, up-to-date formats for the constellation of healthcare stakeholders.

Benefitfocus, the cloud-based benefits management platform and services provider, recently hosted 1,200 benefits professionals at their annual One Place conference in Charleston, S.C. The conference featured updates on Benefitfocus’ strategy, enterprise benefits management technology platform, and partners from its ecosystem; and presented an opportunity to learn from industry thought leaders, technology partners, benefits suppliers, and insurance brokers. On the final day of the event the company facilitated employer efforts to build benefit strategies and experiences at what was billed as the industry’s “largest open enrollment planning event”.

During the event, Benefitfocus updated customers and ecosystem partners on seven key topics, as covered in this blog.

Shift in corporate strategy

Benefitfocus has embarked on a significant strategic realignment. The company is shifting its company strategy from selling software to facilitating a benefits industry platform (or marketplace, such as Amazon). The company has been influenced by the book Platform Revolution, written by MIT professor Geoffrey Parker, who was introduced via a recorded video after having visited the company at its campus headquarters in South Carolina. Parker’s book instructs leaders how to start and run a successful platform business such as Amazon, explaining ways to identify prime markets and monetize networks.

Benefitfocus’ ambition is to “connect benefits buyers and sellers in unprecedented ways” and be accepted in a new bracket of peers, including Amazon, airbnb, and Uber. In practical terms, newly introduced analytics are designed to allow sellers and brokers using Benefitfocus’ SaaS software to segment employer customers and employee populations for “improved benefit strategy, communications and engagement, while giving employers robust visual interactive tools to quantify the value of their benefits programs and serve their employees.”

However, questions regarding the practical ramifications of this strategic shift remained unaddressed in the general sessions, including:

• The shift from a software development culture habituated to a standard, planned software roadmap and update release schedule to a “platform” culture habituated to agile development
• Adaptation of the Benefitfocus sales channel, sales methodology, sales collateral, sales and marketing resource roles, responsibilities and staffing
• Development of an ecosystem partnership within a complex web of coopetition (in which medical carriers, for example, may currently go to market on the Benefitfocus SaaS software, white label Benefitfocus, and/or go to market concurrently with their own home-grown development platforms).
• Development of the benefits administration professional community within Benefitfocus’ ecosystem of employers, consumers, and benefit providers.

Software updates

Benefitfocus platform updates that resonated strongly with benefits partners included:

• Mobile App: It is now possible to email or text health data to a physician, including proof of insurance. This is a service not only for the consumer but for the insurance carrier that wants to have accurate data conveyed to physicians in real-time. The Mobile App aims to simplify consumer engagement, total rewards details, and digital ID cards. Enrollment can now be accomplished using the Mobile App.
• Chatbot: Embedded in AI engine BenefitSAIGE, this 24-hour chatbot drives content and recommendations to consumers every type of benefit at every stage of life. It also frees the HR professional who is ordinarily called to interface with consumers about the benefits platform and benefits companies. Chatbot communications limit delays generated by hand-offs as a consumer inquiry passes to the HR professional, to a benefits broker, to a benefits vendor, and then returns back to the HR professional and finally the consumer. The chatbot also drives appropriate benefits enrollment in “smart moments” that matter to consumers.
• Digital Wallet: This feature enables flexible payment options beyond payroll deduction. Payment using personal credit cards can also be accomplished using the Mobile App. The platform now allows employees to purchase insurance at any time during the year, not just during a two-week open enrollment period.

Other notable added software functionality includes:

• Data interchange and automation enhancements, analytics and communications enabled by AI engine BenefitSAIGE. The AI engine leverages rules-based systems, RPA, machine learning, predictive analytics, and natural language processing. This AI engine aims to improve data interchange, drive insights, improve the consumer experience, and influence transactions during “smart moments”
• Ecosystem productivity enhancements via data exchange, APIs and automation, supported by security and data protection.

Benefitfocus reports that over 25m consumers are now served by its software platform. Clients include 170k+ employers, from Fortune 500 companies to small employers, featuring 17k brokers, 144 medical benefits carriers, and 30+ marquee voluntary and specialty benefit brands.

Data cleansing

The company reports that a $30m investment has produced a dataset with “99.6% data accuracy on first-pass yield, eclipsing the industry average of 95%”. The dataset includes records of 2.7bn data transactions in 2018 alone.

Adding a portable life insurance partner

BenefitsPlace now features Afficiency, an InsurTech that is working with life insurance carriers to offer portable voluntary life insurance benefits.

Adding consumer-directed health partners

The company has also added greater choice of consumer-directed healthcare (CDH) account options, including Wageworks and Payflex. API connections are designed for synchronized, accurate and real-time data exchange. Year-round education and communications should help consumers maximize their CDH contributions, including the triple-tax benefits of funding their HSAs.

Introduction of personal lines insurance products

On the existing software platform, insurance carriers and specialty product suppliers gain a dedicated digital distribution and enrollment channel to more than 23m consumers on the Benefitfocus platform. Carriers included in this first iteration include:

• Bristol West Insurance Group: a member of the Farmers Insurance Group of Companies (PL auto)
• MetLife Auto & Home: Metropolitan Property and Casualty Insurance Company and its subsidiaries, operating collectively under the MetLife Auto & Home brand (PL auto and homeowner)
• Toggle: launched by Farmers Insurance in 2018 (renter’s insurance).

Benefitfocus offers P&C insurance through licensed brokers at discounted rates.

Innovation incubator

Benefitfocus announced its InnovationPlace, a startup partner program. The company aims to introduce innovative products and services to employers and their employees through its SaaS facilitated marketplace. The company has created an innovation incubator on the company’s South Carolina campus, and welcomed its first occupant, Rock Health, an innovator in women’s health.

In this blog, I look at how Sutherland tackled the challenge of health plans maintaining accurate provider data in the state of California.

The challenge: inaccurate health plan data about providers

It’s been difficult for health plans in California to maintain accurate, up-to-date information on the current status of providers in the state. According to outsourcing vendor Sutherland, experience indicates that 60% of provider directories contain serious material errors. Health plan data frequently indicated that doctors were no longer accepting new patients, even though they in fact were. The data frequently presented the state regulatory body, health plans, and patients with inaccurate information about whether doctors continued to practice their specialty, had moved to new locations, or were contracted to work with particular health plans or their products.

The context: gaining access to timely CA medical services

Since 2017, Sutherland has created a shared services model for over a dozen CA health plans that obviates the need for participating California health plans to each separately build and update parallel databases that track the availability of provider appointments for urgent and non-urgent care for health plan members. The State Department of Managed Healthcare (DMHC), which regulates the state’s health plans, requires that health plans and providers make available appointments for urgent and non-urgent care, varying by specialty, from two to 14 days. Until recently, each health plan created and updated its own massive database of providers that participated in each of those plan’s products.

In a state in which Sutherland reports that the average provider contracts with ~ 15 health plan products, the law resulted in a myriad of duplicative efforts, each of which imposed burdensome requirements on providers.

The Sutherland solution

Sutherland has initiated a shared services platform that reduces this burden for health plans, providers, and state agencies, and increases the accuracy of reporting to the DMHC. In particular, Sutherland spearheaded the coopetition of health plans in California in 2017 by creating a shared services model that built and updated the Provider Appointment Availability Survey (PAAS) on behalf of a consortium.

Prior to that, Sutherland had been in conversations with the state of California on a related topic, and that conversation helped initiate Sutherland’s PAAS project with the state. Sutherland had already built a relationship with Blue Shield of CA, which became the anchor client. Other state-based and national health plans joined the consortium in 2017, totaling eight by the end of 2017. By the end of 2018, 12 health plans had joined the consortium and Sutherland now counts that consortium at 14 health plans.

Sutherland estimates that it now touches ~ 100K doctors, each of which has contracts with an average of two plans. This hub-and-spoke shared services model eliminates duplicate outreach to CA providers, saving each participating health plan from the costs of maintaining separate call center facilities and databases, and saving providers from responding to multiplicative health plan outreach regarding the same basic data. Sutherland also manages all the workflows involved with credentialing a new provider, verifying diplomas, board certifications, and combing regulatory authorities for any information on sanctions against providers.

The company estimates that it reduced associated health plan physician data management costs by 75% through elimination of duplicative work and by improvement in survey execution workflow and other improvements. Sutherland estimates that it reduces the touches on providers from 3 to 1 call per practice, improves reporting and other interactions with the California regulatory body, and improves patient access to timely care.

The challenge of rising healthcare costs in the U.S. has been obvious for decades. Or has it? Various interventions have been attempted, but health costs as a percentage of GDP are forecast to continue to climb. National U.S. healthcare expenditure as a percentage of GDP has risen from 17.2% in 2011 to 17.9% in 2017.

In February 2018, the U.S. CMS Office of the Actuary estimated that “growth in national health spending is projected to be faster than projected growth in GDP by 1.0 percentage point over 2017-2026. As a result, the report projects the health share of GDP to rise from 17.9 percent in 2016 to 19.7 percent by 2026.” GDP growth over the last two periods has kept pace with rising healthcare costs over the last two years, but when GDP growth subsides, the healthcare cost challenge will reemerge. The current stalemate at the U.S. federal level about the path forward for healthcare reflects a lack of consensus about root causes and, therefore, advisable policy.

The sector has already undergone major restructuring and intervention, both government and private sector initiatives. This includes:

• The American Recovery and Reinvestment Act of 2009 (ARRA) incentivized adoption of EHRs – the assumption was that a lack of electronic clinical records technology was a primary component of inefficiency and waste. 90%+ of U.S. hospitals have now adopted EHR technology
• The Accountable Care Act (ACA) of 2010 realigned much of American healthcare reimbursement and delivery – the assumption was that decentralized, misaligned organizations created waste and reduced quality. The ACA introduced a raft of initiatives designed to address waste and improve productivity, particularly clinical labor productivity. The results of most of these measures, including the ACA’s Accountable Care Organization initiatives (ACOs) remain inconclusive
• Consolidation: the payer and provider markets have been roiled by restructuring and consolidation. There were 1,412 hospital mergers between 1998 and 2015; physicians also have consolidated into increasingly larger groups. Moreover, the four largest insurers now account for 83 % of the total national market.” [1].

The largest target for improvement in healthcare delivery costs remains the cost of labor. But does more “technology” improve labor productivity? Not necessarily. Technology can drive rather than retard growth in healthcare costs. According to a Health Affairs (HA) article, “technological changes in the [physician and nursing] sector to date have favored, rather than substituted for, those with high skills" [2]. It depends on the type of work or process, on the technology use case, and on the organizational aptitude for adopting new solutions. Administration, management and IT are oft-cited as a source of burgeoning healthcare delivery costs, but these classes of labor may actually be seen as examples to be followed. Over the 15-year period of the HA study, compensation (change in employment x change in earnings) for administration, management and IT rose only 35.3%. Over the same period, compensation for physicians and nurses rose 80.5%.

Taking a step back, have all the industry-level efforts at restructuring mentioned above missed the mark? Have we simply failed to appreciate how unhealthy Americans have become – and therefore overlooked the root cause of precipitous cost increases? The debates and struggles regarding GDP growth, healthcare delivery cost growth, technology adoption, government intervention, and market restructuring may simply be addressing symptoms rather than causes of the rise in U.S. healthcare costs.

The “hidden in plain sight” fact may be that Americans have unhealthy habits which have national ramifications for healthcare costs. In one 2013 study, only 2.7% of the U.S. adult population could be identified with healthy metrics for exercise, diet, smoking, and body fat. As national healthcare expenditures rise towards 20% of GDP, perhaps we should ask whether the challenge of rising healthcare costs can be adequately addressed by industry-level restructuring efforts. Perhaps this challenge can better be addressed by bottom-up rather than top-down initiatives.

[1] The Commonwealth Fund, Insurer Market Power Lowers Prices in Numerous Concentrated Provider Markets, September 6, 2017

[2] Where the Money Goes: The Evolving Expenses of the US Healthcare System, Health Affairs, July 2016