Backlogs have massive costs for government agencies, citizen satisfaction, and business productivity. Here is a holistic, systematic approach to help governments resolve them.
This January, US federal workers returned from the government shutdown facing a daunting task: processing massive new backlogs. The size of the challenge is staggering. Immigration courts canceled over 86,000 hearings during the 35 days of the shutdown and could be playing catch-up for years.1 The Internal Revenue Service received more than 700,000 pieces of mail per day during the latter weeks of the shutdown, and faced millions of unanswered queries from citizens.2
Adding to the challenge is the fact that new backlogs were heaped on top of the many existing backlogs that federal agencies were already working through. Before the shutdown, many immigration courts already faced a two- to three-year backlog.3 The Veterans Administration’s claims backlog, which had reached over 600,000 in 2013, still hovered around 80,000.4 And Freedom of Information Act processing times for many agencies exceeded one year.5
The federal government is not alone in facing down monumental backlogs. State governments, city governments, and judicial courts all struggle with work piling up faster than it can be cleared. And the problem is not limited to the United States. Brazil’s patent office has struggled to shorten a 10-year processing time, and even considered the option of an emergency granting of 230,000 pending filings.6 In India, a 27 million-case backlog has clogged courts for decades—at the current pace, a lawsuit filed today might not be resolved until 2040.7 And in Indonesia, Australia, and South Africa, land claims and registration processes have been backed up for decades.8 The more you look around, the more backlogs seem endemic to government processes everywhere.
Backlogs are certainly not the result of lack of effort. Many government agencies have valiantly tried to reduce their backlogs—often in the wake of bad publicity—but the results rarely stick. That is because attempts to solve backlogs often tackle just the most visible challenges, rather than the underlying causes, and fail to bring about lasting change. At the same time, citizens have become accustomed to extraordinarily fast and targeted service in the private sector, such as highly personalized streaming services, instant-approval loans and credit cards, and before-you-knew-you-needed-them discounts.
So what does lead to sustainable backlog reductions? First, it’s important to understand what a backlog is. While any process may experience a backlog, the government processes whose backlogs draw attention are those that involve systems given individual discrete items to process, often involving a degree of judgment or adjudication. These items may be medical claims, court cases, tax inquiries, security clearances, applications to qualify for services, or a score of other instances. For simplicity we will refer to these collectively as “cases.”
Most backlogs have many causes, but they all share a common strain: They involve case-processing systems that tend to take significantly longer to address cases than to absorb new ones. At its core, every backlog is like the chocolate conveyor belt from “I Love Lucy”: New items are coming faster than workers can remove them.
And just like with the chocolate conveyor belt, the first-place people usually look to solve backlogs is at the process flow, in search of bottlenecks to relieve in order to increase processing speed. It’s a natural place to start and can yield improvements, but when done in isolation, it can shift the problem instead of solving it. A more comprehensive approach is needed—one that addresses the key levers and the key stakeholders common to government backlogs, and that recognizes the connections between people, their processes, and their ecosystems.
The good news is that the controls are in government’s hands. Our research shows that there is a way to help reduce backlogs once and for all—a holistic path that can drive sustainable improvements in government services. It involves these five key levers:
Approaching a backlog-reduction effort by addressing each of these levers can lead to sustainable improvements for agencies, citizens, and budgets. And while addressing these levers collectively can be most effective, thinking through which levers matter most for any specific backlog is a great place to start. But first, we need to step back to understand the value in reducing backlogs.
Agencies often struggle to get the funding needed to fix their backlogs. After all, a backlog is an annoyance, but is it really worth the effort to solve it? The problem with this thinking is it ignores the opportunity costs of a backlog, which can be significant for individuals, communities, and businesses. For example, the US security clearance backlog, which peaked at over 700,000 cases in 2018,9 is a backlog with high opportunity costs. Each clearance case represents an individual who needs access to classified information to do the job right—but instead is unable to do so, or worse, is simply waiting for clearance to be employed. According to a 2018 survey of cleared personnel, jobs that required clearance had an average salary of about US$93,000.10 The downstream effects of the backlog—in employment terms alone—are felt in lost labor market efficiency, forgone income, and reduced tax revenues (not to mention the mission impact of a shortage of qualified and cleared personnel).
Many states face backlogs in everything from human services to examining criminal evidence. With some states facing a serious epidemic of opioid and related drug abuse, a drug-evidence testing backlog can mean delayed justice, which means police could release known drug dealers while they wait on evidence. That means more dealers and traffickers on the street, and more damage to communities.11 The effects on communities can exacerbate backlogs in other state systems—from children in foster care to state and local court systems to elder care.
And government backlogs can reduce the attractiveness of investment and innovation in entire economies. Backlogs in court systems, for example, can deter economic investment by increasing risk, especially for foreign investors, and by enabling anti-competitive behavior, such as bogging down competitors in endless lawsuits or violating agreements with impunity. Backlogs in developing economies in Asia, for example, are soaring, with downstream effects for justice, growth, and long-term development.12 They can harm developed economies too: By one estimate, Italy’s justice backlog reduces GDP growth by 1 percent annually.13
Backlogs can also hinder innovation. Studies by the Center for the Protection of Intellectual Property have found that each year of patent delay can reduce a startup’s employment by 21 percent and sales growth by 28 percent over the five years after approval.14 Patent backlogs can decrease the payoff for R&D, reducing technology progress:15 For example, backlogs in three top patent offices led to more than US$10 billion in reduced global growth each year.16
Backlogs can also reduce citizen satisfaction, and in turn, confidence in government. Trust in government today is at historic lows, with only 18 percent of Americans surveyed saying they trust government to do the right thing all or most of the time.17 For many citizens, case-processing systems are where they encounter government, whether at the registry for motor vehicles, in applying for benefits, or getting permits for their homes or businesses. Long wait times and poor customer experience can further erode confidence in government—no one’s desired outcome.
Faced with overwhelming backlogs and their unfortunate side effects, how do governments respond? All too often, they focus on surface-level problems and use an insufficient toolkit to clear the backlog. Here are the three most common approaches governments try, and why they don’t often work:
So how can government develop strategies that do help? It’s important to understand backlogs within their context—case management processes—and to understand those processes within a context of government and citizen services. Luckily, there’s a clear set of levers that any government agency can follow to tackle their challenges systematically, holistically, and comprehensively.
No amount of process change can make up for government policy that doesn’t get to the heart of the issue, or creates too many unintended consequences. Policy often dictates process, meaning that with the wrong policy, government agencies can achieve technically correct outcomes without solving any real problems. Sometimes policies merely scratch the surface of a problem, merely providing costly band-aids. Sometimes policies are designed inflexibly to prevent abusive behavior, but end up overly restrictive for the average case. Often these policies end up creating processes that have more steps and are far more complex than needed, leading to slower processing, higher backlogs, and lower citizen satisfaction. But finding the right policy, even if it takes a few tries, can yield tangible benefits.
Case processes originate with policymakers who create the service or system, typically by mandating a certain requirement through legislation or executive order. For example, policymakers are ultimately responsible for decisions as broad and significant as having a court system or guaranteeing intellectual property. Policy also drives second-order effects—decisions about what information should be classified, for example, leads both to the need for security clearance processing and influences the number of people who need one. Fixing a backlog starts by understanding how policies shape current outcomes.
Like many court systems, the Hawaiian island of Oahu’s courts faced a backlog on the docket. An analysis revealed that many cases were driven by a small, frequent-offender homeless population, with an average of over 10 cases each. Many of these cases were for violations that are hard for homeless people to avoid, such as sitting or sleeping on sidewalks. The problem wasn’t just that court processes needed to move faster—it was that the court was the wrong venue for many of the cases. Oahu developed a homeless court alternative—the Honolulu Community Outreach Court—that cleared 601 cases in its first year by focusing on solving homelessness rather than on prosecution. Understanding what causes different types of cases—and using policy levers to solve the root problems—is critical to reducing backlogs downstream.
Backlogs never exist in a vacuum. In a system of systems, a small change in procedure here can lead to huge ramifications there. Without examining the upstream and downstream effects, including those outside the case system, well-meaning changes can be counterproductive. That is exactly what happened in the US Department of Justice (DOJ) in its effort to reduce backlogs in immigration courts.
The DOJ’s Board of Immigration Appeals (BIA) functions as an appellate body to the immigration court system.18 It mainly reviews decisions issued by immigration judges, who issue brief oral opinions from the bench immediately or shortly after a hearing. In 1984, the BIA received fewer than 3,000 new appeals and motions each year. In 1994, it received more than 14,000, and in 1998, more than 28,000. To address the large increase and the growing case backlog, the DOJ issued regulations in 1999, and again in 2002, that reduced the demands on BIA reviews of cases. The regulations reduced the number of board members needed to review most cases, made the standard of review more stringent, increased the board’s ability to summarily dismiss cases, and shortened its decision window.
These changes successfully cleared the backlog in a little over a year—but they had unanticipated consequences for other systems. The changes meant that many of the BIA’s decisions, which can be appealed to the federal courts, went from being thorough legal arguments to often merely one-line affirmations of an immigration judge’s decision. The federal circuit courts saw a significant increase in the number of BIA decisions being appealed. At one point, 40 percent of the Ninth Circuit Appellate Court’s pending docket was made up of BIA appeals. The federal court judges became frustrated with the number of appeals and short decisions of immigration judges, and increasingly rejected them. This led immigration judges to take more time when issuing their decisions, which has contributed to a backlog of more than 700,000 cases pending and dockets that in some places go out more than four years. By missing the downstream ramifications of changes to the BIA, policymakers may have ended up exacerbating the problems throughout the immigration justice system.19
Once you get the policy right, think about the customer. Not only is the customer the reason for the system in the first place, but the customer also plays a critical role in determining whether the system succeeds or fails. A system that is well-designed and knows its customers preempts their questions and their needs—with real benefits.
Consider how automated credit card approvals can work so quickly: Analytics capabilities instantly categorize applications to determine if they meet prespecified criteria. Applicants who clearly do are given instant approval, while those who do not are instantly rejected. However, applicants who are anomalous—for instance, those who may merit approval, but don’t fit the standard model—can request to be reviewed by human workers. The system can process millions of low-cognitive load transactions instantly, enabling humans to spend their time on high-value tasks. At its heart, this type of customer segmentation is about deciding which cases merit extra attention and which do not, to allow organizations with constrained resources to prioritize their time effectively.
Segmentation isn’t just about people. The New York City Fire Department (FDNY) has about 300,000 buildings in its inspection universe but only has the capacity to check 50,000 units annually. To better allocate limited resources, FDNY built a system called FireCast to help identify the most at-risk buildings. The system incorporates information gathered during inspections, along with the data sourced from multiple city government departments, and uses the Mayor’s Office of Data Analytics DataBridge infrastructure. The resulting data allows FDNY to segment buildings based on risk of fire. Over time, FDNY has continued to incorporate new data to increase prediction accuracy. Since its launch in 2013, fire department officials report that it has eased workloads by identifying the city’s most fire-prone buildings, some of which hadn’t been inspected in years.20 The backlog doesn’t evaporate, but FDNY gains real insight into what matters most.
Backlogs can create more backlogs. A citizen who enters data correctly, follows the right case management process, and is reassured about processing status and timelines demands limited effort from the workforce. On the other hand, a citizen who doesn’t understand the data entry process, enters the wrong pathway, and frequently calls and emails to check on his or her status ends up generating a multitude of additional tasks for workers. And when there are millions of customers, even five minutes of additional time for each customer means a lot of delay—for 1 million customers, over 40 years of extra work for a case worker!
Bringing a customer experience (CX) approach can help. While most backlog efforts focus on improving internal processes, a CX approach means thinking about how different people might experience the process. It starts with the principles of human-centered design, which focuses on building systems not only for the internal stakeholder, but also for how a customer would encounter the process—and not just “Can we make the information clearer,” but also “What drives someone into my case system?” Building a more thoughtful, fast, and frictionless customer experience—in which the questions and needs of customers are preemptively met—could eliminate an entire layer of citizen response backlogs.21
Thinking through the customer journey is a good place to start. The New York City Department of Transportation (NYCDOT) issues almost 400,000 street excavation permits to construction contractors every year. In 2014, the city replaced its 30-year-old mainframe-based permit management system with a state-of-the-art permitting solution. The new system lets contractors apply for permits 24/7 though any device—desktop, tablet, or smartphone. Once permits are approved, contractors can print them to post at their construction sites. The new system enabled more than 50 percent growth in case capacity, with more than 600,000 permits processed in 2016.22 The changes to customer journey were backed by changes in workforce and process, but the customer never needs to see that.23
And providing a better journey can be about more than just one process—it can be about all customer interactions. After the Veterans Administration backlog came under scrutiny in 2014,24 the VA decided to consolidate its services by launching MyVA, an ambitious program to provide personalized, customer-centric, and timely services to veterans. This system allowed more than 1,000 websites, 956 helpline numbers, 42 call centers, and 220 separate databases to be combined into one customer experience. One month later, the VA launched the beta version of Vets.gov, a single point of contact for all veterans’ information, helping customers to find the right programs and services.25 Since launch, the claims backlog has dropped by more than 80 percent.26
Government agencies exist for a reason—often to provide a service to citizens or other stakeholders. When agencies focus backlog-reductions efforts solely on measuring task processing, they can miss the point—that the number of forms processed isn’t important outside what it means for families served, court cases that reach conclusion, or small business licenses appropriately granted. Outcome-driven cultures focus not on the tasks it takes to complete a process, but instead on meeting the broader goals and vision of the agency or program. An outcome-driven culture still recognizes the need to report on task progress, but it contextualizes such measures as marks toward desired outcomes for those being served.
The workforce is the critical enabler of a backlog-reduction plan. The best-laid plans for tackling a backlog will never survive poor incentives, the wrong training, and a lack of motivation. Connecting work to impact is a critical first step—it helps workers see they’re feeding families, enabling medical care, and helping small businesses, not just filing paperwork.
In 2012, the US Navy faced a massive backlog of requests for verification and reissuance of Navy, Marine Corps, and Coast Guard medals. Military medals have tremendous personal value for veterans and their families. Processing the requests was taking so long that some veterans were not receiving their medals during their lifetimes. The Navy engaged with the National Personnel Records Center (NPRC) to fix the problem. Realizing the potential impact of eliminating the backlog, the small team from the NPRC led a holistic redesign of the records process.
The NPRC team leveraged a number of solutions. They developed a new workflow, identified training gaps among team members, and established milestones and productivity goals. To maintain near-perfect accuracy, the team created a quality and consistency assurance process. Soon, staff identified additional training gaps, flaws in the process, and customer service shortfalls. NPRC iterated on the solution, tweaking the process and training, and integrating a dedicated call center that could provide updates on the status of requests and even reprioritize urgent verification requests.
The team understood that each investigation was not just a records search, but a veteran waiting for help. They set ambitious goals—such as increasing productivity to eight cases per day—and handily beat them, achieving individual productivity of 28 cases per day. In less than three years, the team had provided service to more than 40,000 veterans and their families, with accuracy greater than 99 percent. Motivation alone would not have solved the problem—using multiple levers was critical—but emphasizing the outcomes of the work did help the team pivot when their plans ran into inevitable obstacles.27
Once you have a policy that helps you achieve your goals and you understand who you’re serving, it’s time to dive into the process. That means both analyzing and improving on what’s happening today, and bringing in new technologies when they can offer genuine value.
When tackling a backlog, process engineering—the kind that enabled the first mass-produced automobiles—is often the first approach. Originally developed to optimize the flow of physical objects through a manufacturing facility, process engineering solutions attack backlogs by looking to find bottlenecks or constraints in a process and fixing them. For a government agency, instead of moving a physical product through space, case-processing systems typically move data. That means that the first step in optimizing a process should usually be ensuring flow and access to data throughout the system. And just as the most advanced factories still need humans to manage anomalies and correct errors, government systems need decision rights and information distributed at actionable levels.
Most importantly, addressing bottlenecks is about going beyond the obvious. For example, while the pursuit of greater speed typically looks to reduce steps in a process, sometimes fixing a bottleneck actually means adding steps. The City of New Orleans has faced a serious blight problem, especially after Hurricane Katrina destroyed thousands of homes. The city’s process for addressing blight involves an inspection of a structure, a request to the owner to fix any violations, legal action, and, if needed, a subsequent decision to sell fixable properties or to demolish dangerous ones. The city had one highly experienced inspector who was needed for making the determination to demolish or sell, but one wasn’t enough. In 2014, a substantial backlog of over 1,500 properties awaiting a decision remained, because the inspections and hearings were occurring more rapidly than the experienced inspector could determine what should be done with the buildings.
To address the backlog, the New Orleans Office of Performance and Accountability (OPA) developed a new approach. The team analyzed more than 600 cases using standardized criteria and worked with a data science team to develop an accurate prediction algorithm to classify cases. That helped them develop a scorecard in which inspectors feed attributes of a property and the scorecard utilizes a grading system that produces a “sell score.” This system allowed inspectors with less experience to collect data in the field and receive reliable guidance on whether a property is a good candidate for sale. Then, because the decision to demolish is much more consequential for the property, those cases where “sell” is not the clearly preferred option could be referred to the highly experienced inspector for additional review. The new abatement process adds a step—the secondary review of potential demolition cases—but it enables the most experienced human capital to focus on the hardest cases while allowing more junior inspectors to gain valuable field experience, all while dramatically increasing the volume of cases processed. Once the system was implemented, it eliminated the backlog within three months.28
Most government case systems involve the intake and analysis of data. Over the past decade, there has been a revolution in technologies that collect, analyze, and deliver data, and this offers a real opportunity for government case systems to automate their core processes. Prior to the passage of the Affordable Care Act (ACA), a large part of the verification process for Medicaid eligibility was manual. In Colorado, the process could take up to 45 days before an applicant received confirmation. In anticipation of the expected increases in applications because of ACA, Colorado decided it needed to speed up its processing capabilities.
In 2013, the state embarked on a new strategy for processing its public benefits. It linked disparate data sets across the government to enable faster verification. Then, it developed a self-service portal, the Program Eligibility and Application Kit (PEAK), which automatically determined eligibility for most applicants in real time. Rather than requiring full verification up front, the system approves the most-likely applicants immediately. Here’s how it works:
The enhanced system provides a real-time decision for 80 percent of applicants, reducing the backlog and freeing workers to focus on high-value tasks.29 And other states have followed suit: As of January 2018, thanks in large part to automation, 40 states can make determinations within 24 hours for Medicaid benefit applications.30
Not every hurdle can be eliminated through policy and process flow changes or automation, so organizations need to determine how they will prioritize cases. If not all cases can be processed in a timely fashion, those most critical should be given priority. The US Food and Drug Administration reviews applications for both new medicines and generics. Just a few years ago, the time to review a generic drug was 47 months—nearly four years—from application through approval.31 When generics take so long to reach the market, millions of people are unable to access life-saving medicines, with significant effects on health, family, work, and the cost of social services. Scott Gottlieb, the FDA commissioner at the time, recognized the problem: “The fact is that too many people can’t afford the medicines that they need.”32
So the FDA developed a prioritization plan to ensure it was focusing on the highest value-added applications.33 It chose to prioritize applications that represent the first generics to hit the market—the first three generic versions of a drug create the greatest downward pressure on prices—and it set a goal to bring those applications down to eight months. It combined prioritization with intelligent automation at key points in the process, including robotic process automation to reduce manual data entry.34 The results were significant: Increased competition brought the price of some medicines down by as much as 90 percent and eased access to treatment for millions of people.
Learning and experimenting can be the hardest part of a backlog-reduction plan. You’ve spent a lot of time and effort getting your policies right, learning your customer’s needs, and redesigning your process. To make the process work, you’ve had to retrain employees, incorporate new technologies, and integrate your data. Ideally, you can now sit back and relax as the process flows on autopilot. But of course, it’s never that easy. Pesky anomalies pop up, cases requiring extra attention, cases requiring workarounds. Instead of finding these a source of frustration, you can actually use these—to improve the process, the workforce, and the customer experience.
Most case management systems build a process for the average case or, in the case of a more segmented and tailored process, for several key segments. While commercial processes can choose to not serve customers that don’t fit their processes, government organizations typically have to provide service to all—including one-of-a-kind situations. These exceptional cases may be treated as unwanted annoyances and wind up delayed, ignored or cycling through a set of uncertain decision makers—all of which can further bog down the queue. Exception management seeks to reverse that bias. Once an organization has segmented its cases, automated, and prioritized, what’s left are cases that require investigation, problem-solving, critical thinking, innovation, and improvisation—exactly what humans do better than machines. But addressing those cases is too often relegated to workarounds, sometimes even putting case workers at risk of punishment for violating the process. Moreover, any potential lessons learned from the anomalous cases are usually lost.
Instead of fearing anomalies, organizations that have streamlined their processes can leverage anomalies to learn, grow, and improve. While most new cases simply reinforce existing models, anomalies can help the organization learn, by clarifying the bounds of eligibility or by identifying overlooked populations or emerging problems. And the way that workers address those cases also has value: Workarounds point to places in the process that need to be fixed, potentially by allowing greater improvisation or by devolving decision rights. Capturing the lessons learned from anomalous cases can then help train workers for the unpredictable—or even demonstrate that those cases are not so anomalous after all, but are consistently being underserved.35
Most backlogs don’t have easy, one-step solutions—that’s part of why they are so persistent. A key part of understanding what causes the backlog, and what solutions can ameliorate it, is experimentation. Sometimes what seems like a fix at step 1 just exacerbates a problem at step 17. That is why experimentation is a process, not a one-time fix. It means piloting ideas, tracking their performance, learning, testing in new environments, and scaling what works. Often the best place to start is by asking the stakeholders themselves.
Courts in the Philippines faced mounting backlogs for decades. Attempts at reform had failed. Then in 2011, they tried a new approach. A pilot program in Quezon City, which had more than 40,000 cases pending, began by assembling stakeholders from across the court system, including officials, judges, prosecutors, public defenders, and private attorneys. The participants proposed a series of solutions, from changing timelines and filing procedures to offering off-ramps from the court system for some cases.36 The pilot incorporated these inputs and designed a new set of procedures to speed court cases, then worked with attorneys practicing in the system to get their buy-in.37
The pilot programs learned, kept what worked, and expanded. Within two years, three pilots had changed how Quezon City courts operated. The new procedures, developed with stakeholder participation, had been brought to other courts facing backlogs, speeding processing times. Another pilot dealt a quick blow to over 9,000 of the oldest cases on the docket, clearing nearly 30 percent of the backlog. And an eCourt program introduced an automated case management system.38 Within five years, the Philippines had introduced eCourt processes to nearly 300 courts across the country.39
Reducing backlogs isn’t easy, but it is achievable. First, it’s important to bring in different viewpoints. Policymakers, agency leads, case workers, and customers all merit a seat at the table, because each plays an important role in fixing—or exacerbating—a backlog. Start by convening people from all levels in an open, safe environment. Much of the data, expertise, and ideas you need may already be within the organization. For example, a seasoned workforce can often point to where and under what conditions case processing slows down, what the risks and rewards are for how they allocate their time, and when to circumvent the system to get things done. These insights can help sort through mounds of data that may be underutilized.
Next, realize that whenever stakeholders point to workarounds needed to “get things done,” you have a broken system. Every such workaround is evidence that the current system doesn’t allow you to get things done. Document what those things are and where they happen. That can help you build the shadow process map—the flow of what really happens as cases are processed.
No one can wave a wand and create a better policy, a cleaner process, or a more productive workforce culture. But every organization can pilot new ideas. After assessing the process and the shadow process, leverage your data to understand which levers to target first, focusing on highest-value adds. Before starting a pilot, be sure to understand the perspectives of critical stakeholders, and take time to incorporate their feedback.
Finally, many governments, at national, state, and city levels, have developed in-house digital expertise. Leveraging these digital organizations can help frame the art of the possible, early and often, for government agencies.
As digital technologies mature, we wait in fewer lines than ever before, both literally and figuratively. Government agencies are adopting measures to make sure the same is true for the citizen experience. By approaching backlogs holistically, armed with the proven levers outlined here, government leaders can make real improvements for their mission, for their workforce, and for the citizens they serve.