Key Takeaways
Make government credible by fixing the services people actually depend on.
Service is the test. Charlie Baker and Steve Kadish argue that public trust is built less by political victories than by whether government can deliver health coverage, reliable transit, child protection, and other essential services. Their method grew from three decades of public and private sector work, then faced its hardest test when Covid-19 reached Massachusetts in 2020. The state had to act while testing was scarce, evidence was changing, and the costs of both action and delay were immense.
Results are democratic work. The authors’ central claim is that delivery is not administrative housekeeping after the “real” political work. For people who depend on a public service, implementation is the policy they experience. Government also cannot simply abandon services that are difficult or unprofitable: often there is no substitute provider.
What's striking is the book's insistence that democratic legitimacy is partly built in ordinary transactions: a call answered, a benefit processed, a train that arrives. Political theory often focuses on consent, representation, and rights; service quality adds a daily, practical test of whether institutions deserve confidence. The framing is persuasive, but it should not turn trust into a popularity score. A necessary public decision can be unpopular, and an efficient service can still be unfair. Credibility depends on explaining tradeoffs and treating people justly, not only on making systems run smoothly.
Choose leaders for execution, then give them a real team.
Capability comes first. The authors use “people are policy” to mean that policy only becomes real through the people who design, manage, and deliver it. Baker chose Stephanie Pollack, a prominent transit advocate from the opposing party, to lead transportation because she combined deep subject knowledge with the appetite to tackle entrenched problems. Their later work at the MBTA also showed why a single charismatic leader is insufficient: the agency needed distinct leaders for overall management, finances, and daily operations.
Build capacity deliberately. Look for knowledge, practical know-how, collaboration, and varied experience. Mix strong incumbents with new hires, define responsibilities, and add a chief operating officer or a dedicated team when delivery is nobody’s clear job. For a crisis initiative, “dedicated” means people have substantial time assigned to it, not merely good intentions alongside a full-time workload.
This is an organizational-design argument as much as a hiring lesson. Teams work better when complementary skills are explicit: technical expertise identifies sound options, while operational expertise makes those options usable. The choice of Pollack also illustrates how a leader can treat disagreement as a source of competence rather than a loyalty test. Yet “the right people” can become a convenient explanation for failures caused by inadequate authority, funding, or policy. Hiring matters most when leaders also give staff the means and permission to act.
Pair hard numbers with the human stories hidden inside them.
Two kinds of evidence. “Data evidence” means information about a program’s operations, finances, staffing, rules, and performance. “Points of pain” are specific accounts of how a problem affects people. The authors argue that neither is enough alone: statistics reveal scale and patterns, while personal experiences show what those patterns mean and where a process breaks.
Rural internet made the gap visible. In 2016, nine of 53 targeted western Massachusetts towns had limited service and 44 had none, despite a decade-old state initiative. Residents described businesses losing customers and children struggling to do homework. Reviewing town-by-town conditions showed that a single model would not fit different geographies and finances. The state offered five implementation models. By fall 2021, 50 towns had new service, with the remaining three in progress.
The pairing resembles mixed-methods research: quantitative data estimates the pattern, while interviews help explain its mechanisms and consequences. It also disciplines empathy. A moving anecdote can reveal an overlooked failure, but it cannot show how common that failure is; aggregate data can show prevalence while concealing who bears the burden. The strongest diagnosis triangulates both, then checks whether the proposed remedy changes the lived experience. Stories should open inquiry, not substitute for representative evidence or decide policy by emotional force alone.
Treat implementation as the work, not the footnote to policy.
A promise needs a route. The authors distinguish what to do, such as a law, service, or program, from how to do it: the people, systems, decisions, and day-to-day work that make it function. The “work-management triangle” describes the constraints that must be balanced: time, scope, and resources. The authors prefer setting an early deadline, then narrowing the work to what can be delivered and improved in stages.
Speed can create traction. In spring 2020, Massachusetts set a two-week target for a statewide contact-tracing operation. A team divided the work into streams for staffing, protocols, technology, data, and communications. First calls began after 12 days; the operation soon employed more than 1,000 callers. The early version was not perfect, but a working service allowed the team to learn and improve while the need was urgent.
This approach has a clear connection to iterative product development: deliver a usable version, observe what happens, and refine it rather than betting everything on a distant launch. In government, staged delivery can make public commitments more credible because problems surface before they become system-wide. But urgency is not automatically beneficial. A deadline can focus attention, or it can encourage corner-cutting and exhaust staff. The useful discipline is not speed for its own sake; it is making the deadline visible while preserving testing, legal review, and a way to correct mistakes.
Design cooperation around who must agree, not who has authority.
Public work crosses boundaries. Government decisions involve legislatures, agencies, unions, contractors, communities, and courts, each with different responsibilities. The authors emphasize trust, humility, respect, and compromise because an executive order cannot make every partner cooperate. Their “round the table” technique asks each participant, regardless of rank, to share concerns and relevant information in turn.
Bridgewater required a coalition. At a state facility for men with severe mental illness, security practices had displaced meaningful treatment. Families and advocates described the harm; peer comparisons showed the facility was an outlier. The revised model separated correctional security duties from mental health care. Union negotiations helped move staff into other jobs, health professionals expanded services, and legislators provided funds. Reported seclusion fell by 99 percent and physical restraints by 98 percent.
The case shows that collaboration is not merely a pleasant leadership style. It is a practical response to fragmented authority: each participant controls something the solution needs. The round-the-table format can surface knowledge that hierarchy suppresses, especially from frontline staff. Still, a meeting does not erase unequal power or guarantee that every voice carries equal weight. Good collaboration needs clear decision rights, follow-through, and attention to whose interests are easiest to overlook. Compromise is productive when it changes the conditions people face, not when it simply quiets conflict.
Measure individual experiences; averages can declare victory while people wait.
Disaggregate the result. Averages can hide severe outliers. At the Massachusetts Department of Children and Families, the average social-worker caseload was 19, close enough to the target of 15 to appear manageable. But more than 40 percent of workers carried 20 or more cases. Looking office by office and worker by worker revealed where support was most needed.
Change the conditions, then check again. The department added social workers and support staff, restored management capacity, and supplied help with transportation, childcare, and clinical advice. By August 2017, the share of workers with 20 or more cases had fallen from 42 percent to 11 percent. The median caseload dropped from 19 to 16. The goal was not a better statistic for its own sake, but more attention for children and families.
This is a lesson in distribution, not merely measurement. A mean can improve while the people with the worst experiences remain stuck, a problem familiar in public health and education as well as customer service. Reporting medians, ranges, and subgroups makes inequities harder to hide and interventions easier to target. Yet granular metrics can also create surveillance or encourage staff to optimize what is counted. In child welfare, a caseload number cannot capture case complexity or safety. Measures need frontline interpretation and a qualitative check on whether the underlying service improved.
Reset failing systems with focused leadership, not another report.
A diagnosis must lead somewhere. Snowmageddon, with 94 inches of snow falling in Boston in 30 days, shut down the MBTA and exposed years of neglected maintenance and weak management. Earlier reports had not produced sufficient change. A new six-member panel reviewed costs, infrastructure, governance, employee practices, and customer experience, then issued recommendations in about 40 days. Its findings included a $7 billion need just to maintain existing infrastructure and the fact that the agency had spent only $2.3 billion of $4.5 billion authorized for capital work over five years.
Change accountability and capacity. The state created a dedicated oversight board that met weekly, recruited new senior managers, accelerated hiring, and focused first on customer service and financial control. Within two years, state-of-good-repair spending nearly doubled. The turnaround remained unfinished, but the agency was better staffed and able to move projects forward.
The case distinguishes diagnosis from institutional change. Reports can name a problem without altering who decides, who is accountable, or whether the organization has people capable of acting. A dedicated oversight body and faster hiring changed those conditions. The model is especially useful when authority is diffuse, but it carries a risk: emergency governance can centralize power without ensuring durable public accountability. A strong reset needs a defined mandate, transparent decisions, and an exit or review point so temporary intervention does not become unexamined control.
Make technology succeed by fixing the operation around it.
Software is part of a service. The Massachusetts Health Connector had two failed enrollment periods, a call center abandoning up to 40 percent of calls, and information systems that did not reliably connect premiums, eligibility, and coverage. A software fix alone could not repair broken handoffs or give callers accurate answers. New executive and operations leaders, a restructured board, active consultant management, and a 45-day recovery plan reset the work.
Test the whole chain. The team fixed urgent customer cases, increased staffing and training, corrected data exchanges, and released system improvements in short cycles. By the second week of April, the abandoned-call rate was below 3 percent and the average wait to answer was under a minute. In 2019, the Connector enrolled 65,000 new members and reached its largest membership to that point.
This is a sociotechnical view of technology: software, rules, staff, vendors, and user behavior form one system. A failure at any handoff can make a technically functioning application useless to the person seeking coverage. The authors’ insistence on production-like testing is valuable because realistic volume and imperfect data expose weaknesses that idealized demonstrations miss. The deeper lesson is also about procurement and power. Vendors can supply expertise, but public agencies remain responsible for defining outcomes, managing contracts, and protecting customers when systems fail.
In crisis, act on today's evidence and revise tomorrow's plan.
Uncertainty does not pause the work. During Covid-19, Massachusetts built a command center within 48 hours, gathered medical advice, and tracked cases, hospital capacity, and testing while evidence changed rapidly. Surveillance testing at one nursing home found 77 positive residents among 91, including people without symptoms, prompting a stronger focus on long-term care. The response also included expanded testing, contact tracing, and support for food and housing needs.
Adaptation is part of execution. The initial vaccine appointment website failed under heavy demand. The state expanded its technology team, repaired capacity, and introduced preregistration. The vaccination strategy also shifted from mostly local delivery toward mass sites, pharmacies, hospitals, and community health centers. By mid-April 2021, Massachusetts ranked near the top among states in the share of residents vaccinated. The crisis still brought profound losses and preventable failures.
The strongest feature of this account is not a claim of control, but the use of feedback when control is impossible. A crisis response is a sequence of decisions made with incomplete information, so it needs public metrics, channels for bad news, and the humility to revise. Those same powers require scrutiny: emergency measures affect livelihoods, liberties, and unequal communities. Transparent reasoning and disaggregated outcomes help distinguish a defensible precaution from an unexamined habit. Adaptability supports legitimacy only when the public can see what changed, why, and with what consequences.
Analysis
Results is best read as a field guide to the gap between public intention and public experience. Its central contribution is not simply that people, evidence, implementation, and measurement matter. Most management books say some version of that. The useful move is to bind them into a repeating operating discipline and apply it to systems where authority is divided, failure is visible, and customers often cannot choose another provider.
The cases make execution tangible. Broadband improves when local communities can choose among service models. The Health Connector recovers when technology fixes are joined to staffing, vendor oversight, and accurate customer records. The MBTA moves beyond repeated diagnosis only when governance, hiring, and capital delivery change together. Across these examples, the framework treats organizational performance as a designed capability, not a mood or slogan.
Its democratic argument is equally important. Trust is not restored by asking the public to admire government in the abstract. People encounter government as a parent seeking help for a child, a rider waiting for a bus, or a patient checking insurance. Reliable service can make institutional competence visible in daily life. Yet competence is not the same as justice. A program can process cases quickly while excluding people through its eligibility rules, or meet an average target while leaving a minority with intolerable delays. The book's attention to human experience offers a corrective, but equitable outcomes must be specified and measured, not presumed.
The evidence also has limits. These are retrospective accounts written by leaders describing their own administration, with strong examples of improvement but no systematic comparison to similar initiatives that did not use the framework. Selection effects and political self-presentation are unavoidable. The framework should therefore be treated as a practical diagnostic, not a proven universal formula. Its most durable lesson is methodological: expose the assumptions, listen beyond the leadership circle, define who owns each decision, track effects at the level people experience, and revise openly when the evidence contradicts the plan.
People Also Read
Glossary
Results Framework
A delivery methodA four-part method for moving from a public problem to sustained improvement. It begins by assembling capable people, then defines the problem through evidence and affected people’s experiences, develops both a solution and its implementation plan, and repeatedly measures performance to make adjustments. The parts are interdependent rather than a one-time checklist.
People are policy
People turn policy into serviceThe principle that policies and programs become real only through the leaders and teams who design, manage, and deliver them. It directs attention to the skills, experience, values, collaboration, and staffing needed for implementation. It also means that building organizational capacity is part of making policy, not a separate personnel task.
Points of pain
Specific human impactsConcrete accounts of how a policy, service, or operational failure affects people. These may come from customers, families, staff, advocates, or service providers. Paired with quantitative and operational evidence, points of pain reveal consequences that totals and averages can miss and help define what a successful repair should change.
Data evidence
Operational facts and trendsThe quantitative and qualitative information used to understand a program’s condition. It can include staffing, costs, rules, transactions, service processes, customer outcomes, historical patterns, and comparisons with similar organizations. The authors emphasize analyzing evidence at multiple levels, from individual cases to overall program trends.
work-management triangle
Time, scope, and resourcesA planning model that treats project time, the work to be completed, and the resources available as connected constraints. Changing one affects the others. The authors recommend setting an early timeframe to create urgency, then defining a realistic scope and securing the people, money, systems, and partners needed to deliver it.
Checklist to Green
A project readiness reviewA ten-point review used to determine whether a project is ready to launch or needs help. It checks staffing, evidence, customer impact, goals, governance, project management, deadlines, communications, and performance measures. Teams can also use it to diagnose stalled work and identify the missing conditions needed to get progress back on track.
round the table
Structured group listeningA facilitated discussion in which each participant is invited to speak, regardless of title or position. It is used to surface concerns, facts, and ideas that ordinary status meetings may suppress. The leader often speaks first, and the group records key issues and next steps so that speaking openly leads to action.
Download PDF
Download EPUB
.epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.