Topic 2.1 — The Public Value Scorecard and Outcomes Frameworks

Why this matters in public value

Most public and social-sector leaders can produce a strategy document that speaks fluently about outcomes. Far fewer can produce, a year later, the evidence that shows whether those outcomes actually moved. The gap between the two is not usually a shortage of data — most organizations collect enormous volumes of it — but a shortage of a disciplined structure that turns raw activity counts into a defensible claim about value created. Topic 1.2 — The Strategic Triangle: Legitimacy, Value, and Capacity set out legitimacy and support, the value proposition, and operational capacity as the three legs an organization's strategy must satisfy at once (see Topic 1.2 — The Strategic Triangle: Legitimacy, Value, and Capacity). This topic is about the tools that make those three legs measurable rather than merely asserted.

The stakes are immediate and recurring. A funder or legislature asks, at the next budget round, "what did we get for this money" — and an organization with only output counts (clinics opened, cases processed, training sessions run) has no answer to the outcome question underneath it. A program team believes it is succeeding because activity is high and busy, without having separated activity from the change in the world the activity was meant to cause. A board approves a new initiative on the strength of a persuasive narrative, then has no agreed basis a year later for deciding whether to continue, scale, or stop it. Each of these is a measurement failure with strategic consequences, not a technical inconvenience.

The tools in this topic — Mark Moore's Public Value Scorecard, the logic model, and results-based accountability and its relatives — do not solve every measurement problem a public or social-sector organization faces. Later topics cover the harder, more technical layers: monetizing outcomes through cost-benefit and cost-effectiveness analysis (see Topic 2.2 — Social Cost-Benefit and Cost-Effectiveness Analysis), valuing outcomes that resist monetization through social return on investment (see Topic 2.3 — Social Return on Investment and Impact Measurement), and establishing that a program actually caused an observed change through formal program evaluation (see Topic 2.4 — Public Sector Econometrics and Program Evaluation). What this topic supplies is the layer underneath all three: a shared vocabulary and a small number of durable frameworks for stating, before any of that technical work begins, what an organization is actually trying to change, for whom, and how it will know.

Core concepts

The Public Value Scorecard is Mark H. Moore's proposal for tracking each leg of the strategic triangle with its own set of indicators, rather than defaulting to the financial and operational measures a private-sector balanced scorecard would use. Moore adapted the balanced scorecard — Robert Kaplan and David Norton's tool for tracking a business across financial, customer, internal-process, and learning perspectives — because a single financial bottom line cannot represent an organization whose central claim to exist is that it creates value no market price captures. The Public Value Scorecard instead asks an organization to specify measures under three headings that mirror the triangle: a legitimacy-and-support account (is the authorizing environment satisfied — funders, overseers, the public), a public-value account (is the value proposition being realized — outcomes for citizens and clients), and an operational-capacity account (is the organization able to sustain and improve delivery — staff, systems, finances, partnerships). The discipline is in refusing to let any one account stand in for the whole; an organization that reports only financial efficiency, or only client satisfaction, has not reported against the triangle at all.

Logic models give the value account of the scorecard its internal structure. A logic model lays out the causal chain a program is claiming to run: inputs (the money, staff, and other resources committed), activities (what the organization does with them), outputs (the immediate, countable products of those activities — sessions delivered, permits issued, households visited), outcomes (the changes in knowledge, behavior, condition, or status that outputs are meant to bring about, typically over months or a few years), and impact (the longer-term, often population-level change the outcomes are meant to contribute to, usually shared with other actors and harder to attribute to any one program). The distinction between output and outcome is the single most consequential one in this topic: an output is something the organization did; an outcome is something that changed for someone else as a result. A training program's output is the number of people trained; its outcome is what those people can then do, or how their circumstances change, that they could not before.

Theory of change is the closely related discipline of making the assumptions behind a logic model explicit rather than implicit — why the organization believes activity A will actually produce outcome B, and under what conditions that causal claim might fail. A theory of change is not simply a more detailed logic model; it foregrounds the "if... then... because" reasoning and the contextual assumptions a logic model's boxes-and-arrows diagram tends to leave unstated. Topic 2.3 uses theory of change as the starting point for impact measurement in the social sector in more depth (see Topic 2.3 — Social Return on Investment and Impact Measurement); here, it matters mainly as the discipline that keeps a logic model honest about what it is actually claiming to cause, as distinct from what it merely hopes follows.

Results-based accountability (RBA) is the best-known member of a family of outcomes-based frameworks used across governments and the social sector to connect population-level outcomes to the performance of specific programs. Developed by Mark Friedman and set out in his book Trying Hard Is Not Good Enough, RBA separates two distinct kinds of accountability that organizations routinely conflate. Population accountability asks how a whole community or jurisdiction is faring on a condition of well-being — child poverty, community safety, educational attainment — using a small number of headline indicators that no single agency owns or fully controls. Performance accountability asks how well a specific program or agency is doing, using three kinds of performance indicator Friedman's framework insists on distinguishing: how much was done (effort — the output count), how well it was done (quality), and whether anyone is better off (effect — the outcome). RBA's practical method, sometimes called "turn the curve" analysis, works backwards from a chosen indicator — asking what story the data tells, what is causing that story, who the partners are with a role in changing it, what would work, and what the organization proposes to do — a structured version of the reasoning a logic model and a theory of change formalize separately. RBA is one instance of a wider pattern of results-based management, closely related to the routine practice of monitoring and evaluation used across international development and public administration, promoted in various forms by national governments, philanthropic funders, and multilateral bodies as a shared discipline for connecting spending to outcomes rather than only to outputs.

Outputs, outcomes, and the scorecard together. The reason these frameworks belong in one topic, rather than three, is that they answer the same underlying question at different points in an organization's planning and reporting cycle. The logic model is drawn at design time, to state the causal claim. Results-based accountability and comparable frameworks are applied at reporting time, to keep performance measures honest about the difference between effort and effect. The Public Value Scorecard sits above both, as the standing structure that ensures outcome evidence — however it was produced — actually reaches the legitimacy and capacity conversations where budget and mandate decisions get made, rather than staying inside a program team's own reporting. Together, the three frameworks are what turn public value from a claim an organization makes about itself into a claim it can defend with evidence.

Best practices

Draw the logic model before the indicators, not after. Choosing what to measure before agreeing what you are claiming to cause produces a list of whatever is easy to count; draw the inputs-to-impact chain first, then pick measures that test each link in it.
Report against all three scorecard accounts together, not one at a time. A legitimacy update with no outcome evidence, or an outcome report with no capacity account of what it cost to sustain, each tell only part of the story a funder or overseer needs to make a real decision.
Separate effort, quality, and effect explicitly in every performance report. Borrow Friedman's three-way split even outside a formal RBA implementation; a report that blends "sessions delivered" with "lives improved" without distinguishing them will always default to whichever is easier to inflate.
Choose a small number of headline population indicators and hold them for years, not quarters. Population-level conditions move slowly and are affected by many actors; changing the headline measure every reporting cycle makes it impossible to tell whether the curve is actually turning.
Name, out loud, what the organization does not control. A logic model that implies a single program fully causes a population-level outcome invites exactly the skepticism that undermines legitimacy when the outcome does not move on schedule; state the other contributing factors from the start.
Publish the outcomes framework before the budget round, not after. An outcomes framework produced to justify a budget decision already made reads as post hoc rationalization; one published in advance of the round is evidence a real decision is being tested against it.
Keep the scorecard and the logic model living documents, revisited when context changes, not annual artefacts. A logic model drawn for one policy environment can quietly stop matching reality — a new law, a demographic shift, a funding change — long before a scheduled review would catch it; treat both as owned by delivery teams, not a compliance exercise for a central office.
Keep the performance-account measures close to the people who can act on them. A program team can improve quality and effort measures directly; hand them population-level outcome targets they cannot move alone, and the measurement becomes a source of demoralization rather than direction.
Treat a missed outcome target as a prompt to ask why, not a verdict to hide. The "turn the curve" habit of asking what is causing the trend before assigning blame keeps outcomes reporting useful; a culture that punishes missed targets teaches staff to protect the number rather than the outcome.
Make the assumptions in the theory of change falsifiable. An assumption written so vaguely that no evidence could ever contradict it is not adding rigor to the logic model; write assumptions specifically enough that a future evaluation could show them to be wrong.

Questions to discuss with your team

Can we currently distinguish, for our most important program, an output from an outcome — and do our own staff use those two words correctly? Many organizations report activity counts using the language of outcomes without noticing the substitution, which quietly erodes the difference this topic treats as foundational. Use this question to pull an actual current report and check, line by line, which numbers describe what the organization did versus what changed for someone else. Discuss where the language has drifted, and whether that drift reflects genuine confusion about the program's logic or simply that outcome data is harder to get than output data. If the honest answer is the latter, that is itself a finding worth recording, not a reason to keep using output language dressed as outcomes.

Where does our current reporting collapse effort, quality, and effect into a single number, and what does that number actually tell a reader? Friedman's three-way split exists because organizations routinely default to reporting effort — the easiest of the three to count — while implying they have reported effect. Use this question to identify one flagship metric the organization currently reports and work out, as a group, which of the three it actually measures. Discuss what it would take to report all three separately for that metric, and whether doing so would change how the organization is perceived by the people who currently only see the blended number.

If our headline population outcome does not move for two more reporting cycles, what would we do differently, and what would we not change? This question tests whether the organization is genuinely using its outcomes framework to steer, or simply producing it as a reporting obligation. Use it to distinguish which parts of the current strategy are contingent on the outcome data and which are fixed regardless of what the data shows. Discuss honestly whether any part of the current program would survive unchanged even if the evidence turned firmly against it, since that is the part not actually being managed by the framework.

Does our Public Value Scorecard, if we have one, report all three legs of the triangle, or has it quietly become a scorecard for one leg only? It is common for an outcomes framework to start with genuine three-leg ambition and drift, under reporting pressure, into an outcomes-only or a finance-only document. Use this question to audit the last two or three formal reports against the three-account structure this topic describes. Discuss which leg tends to get dropped first under time pressure, and what that says about which leg the organization implicitly treats as optional.

Who outside this organization would need to agree with our logic model for it to be credible, and have we actually asked them? A logic model drawn entirely by the program team that designed the intervention can quietly encode optimistic assumptions nobody outside the room is positioned to challenge. Use this question to identify which external actors — funders, delivery partners, the people the program serves — have a stake in whether the causal chain is right, and whether any of them have actually reviewed it. Discuss what changed, if anything, the last time an outside perspective was invited into that review.

What is the smallest set of population and performance indicators we could report that would still let an outsider judge whether this organization is creating public value? Measurement systems tend to accumulate indicators over time far faster than they retire them, until the framework becomes too large for anyone to actually use in a decision. Use this question to attempt, as an exercise, to cut the current indicator set by half without losing the ability to answer the population-accountability and performance-accountability questions this topic describes. Discuss what resistance that exercise surfaces, since resistance to cutting a specific indicator is often a sign of what the organization is really being held accountable for, whether or not that matches its stated strategy.

In practice: a public value example

A regional network of family-support charities — fictional, illustrative of a pattern rather than a report on any real organization — had spent a decade running home-visiting programs for new parents in low-income neighborhoods. Its annual reports to funders were detailed and, by output measures, impressive: thousands of home visits completed each year, high attendance at parenting workshops, strong satisfaction scores from families who took part. On the strength of that record, the network had grown steadily, adding new local branches roughly every two years.

A new national funder, considering a multi-year grant large enough to double the network's reach, asked a question the network's own board had never formally put to management: what evidence showed that children in families who received the home visits were actually doing better, later, than comparable children whose families had not. The network's leadership realized, working through the question, that its reporting had never separated effort from effect. Visit counts and satisfaction scores were genuine evidence that the program was being delivered and was well received; neither was evidence that it changed a child's developmental outcomes, which was the claim embedded in the network's own mission statement and, implicitly, in every grant application it had ever written.

The network built a logic model from scratch, starting from the outcomes it actually wanted — improved early child development indicators, reduced instances of substantiated child neglect, stronger parental confidence — and worked backwards to the activities and outputs each was meant to depend on, naming the assumptions in between explicitly. It adopted a simplified results-based accountability structure: population indicators tracked at neighborhood level and shared with local schools and health services, alongside performance indicators its own teams could act on directly — visit frequency, workshop completion, and staff-assessed parenting confidence at three and twelve months. Two of the five outcome indicators did not move as hoped in one neighborhood in the first reporting cycle. Rather than suppressing the result, the network used its own "turn the curve" process to investigate, found that a change in local health-visitor staffing had disrupted referral pathways, and adjusted its partnership arrangements rather than its core model. The funder renewed and expanded the grant, citing the honesty of the reporting, not the absence of a shortfall, as the reason for confidence.

Four sector lenses

Local government

A local authority's outcomes framework typically sits inside a wider set of nationally defined indicators it did not choose but must report against, alongside locally chosen measures that reflect its own council's priorities. This dual structure means a local scorecard often has to reconcile two audiences with different questions — a national inspector asking about compliance with a standard indicator set, and local residents asking about a specific, visible local condition, such as a particular street or service. Logic models at local-government scale are usually easiest to build for a single service (a recycling collection change, a library redesign) and hardest to build credibly for cross-cutting outcomes such as community cohesion, where many actors outside the council's control also contribute.

National government

National outcomes frameworks operate at a scale where the population-accountability and performance-accountability distinction from results-based accountability becomes essential rather than optional: a national health or education indicator is affected by hundreds of programs, only a few of which any one department funds directly, so a single ministry claiming sole credit or sole blame for a national trend is rarely defensible. National governments are also the most common source of large, centrally mandated outcomes frameworks that delivery bodies further down the chain must adopt, which creates the recurring risk that the framework is designed for aggregation and comparison across departments rather than for actually steering any one program's delivery.

Nonprofits and non-governmental organizations (NGOs) more often build their own logic models and theories of change from first principles than adopt a framework handed down by an authorizer, because their funders are frequently the ones asking for the causal case rather than supplying a standard template. This gives social-sector organizations more freedom to design an outcomes framework that fits their actual program logic, but also more exposure to funder-driven measurement fatigue, where each grant carries its own reporting requirements and an organization ends up maintaining several parallel outcomes frameworks for the same underlying work. Results-based accountability's population-versus-performance split is particularly useful here, since it gives a nonprofit language to tell a funder plainly that it can be held accountable for its own performance measures but not, alone, for a population indicator many other actors also influence.

Multilateral and international

Multilateral bodies operate results-based management at a scale where the logic model's assumptions travel across dozens of national contexts at once, so an intervention logic that holds in one country can fail silently in another without the headline global indicator revealing which context is the problem. Results frameworks in this setting are often built to satisfy multiple donor governments simultaneously, each with its own reporting preferences, which pushes towards standardized indicator sets (of the kind promoted through international results-based management guidance) at the cost of local specificity. The gap between an output an implementing partner can report quickly — trainings delivered, clinics supplied — and an outcome that only shows up in a national survey years later is at its widest here, making the discipline of stating the assumed causal chain explicitly, rather than assuming it, especially important.

Common failure modes

Output masquerading as outcome. An organization reports activity counts using outcome language — "improved community safety" when the actual measure is patrols conducted — either from genuine conflation or because outcome data is harder and slower to collect. The cost surfaces when an external reviewer asks for the outcome evidence directly and finds only outputs behind the claim, damaging the organization's credibility more than a modest, honestly reported outcome would have.

Indicator proliferation. A scorecard or logic model accumulates measures faster than it retires them, driven by each new stakeholder request adding one more line rather than replacing an existing one. The eventual result is a framework too large for any manager to actually use in a decision, which quietly reverts the organization to informal judgment while the official framework becomes a reporting artefact nobody consults.

Target capture. A performance measure, once attached to funding or reward, becomes the thing staff optimize directly rather than the outcome it was meant to proxy for — the pattern Goodhart's law describes, where a measure used as a target stops being a reliable measure. This topic's home for that dynamic in the public sector specifically, including how quality regimes try to detect and limit it, is elsewhere (see Topic 3.11 — Quality, Safety, and Performance Management); here it matters as a reason to keep effort, quality, and effect measures separate and to revisit targets before they distort behavior.

Attribution overreach. A program's reporting implies sole causal responsibility for a population-level outcome that many other actors also influence, either through careless language or genuine belief. When the outcome fails to move — often for reasons outside the program's control — the organization absorbs blame disproportionate to its actual contribution, which then discourages honest reporting of shortfalls in future cycles.

Framework without feedback. An organization builds a genuinely well-designed logic model and outcomes framework, then never revisits it against actual results, so it hardens into a fixed document rather than a live management tool. The strategy underneath it keeps running on the original assumptions years after evidence would have justified revising them, at the cost of the very adaptiveness the framework was meant to enable.

Maturity model

Sub-dimension	Initiate	Develop	Standardize	Manage	Orchestrate
Logic models exist for key programs	Causal logic understood informally by program staff, not written down	Some programs have a written logic model, inconsistent depth	Every major program has a logic model built to a common template	Logic models reviewed and revised against actual evidence on a set cycle	Logic models shared and jointly revised with delivery partners and funders
Outputs and outcomes are reported separately	Reporting blends effort, quality, and effect into single headline numbers	Some reports separate them, but inconsistently across teams	Standard reporting format enforces the effort/quality/effect split	Outcome shortfalls trigger a defined "why" investigation, not just a note	Effect data shared publicly and used jointly with partners to steer resourcing
The Public Value Scorecard covers all three legs	No formal scorecard; reporting is ad hoc and usually output-only	A scorecard exists but one leg (often capacity) is thin or missing	All three legs reported together on a fixed cycle	Scorecard results feed budget and mandate decisions directly	Scorecard co-developed with authorizers and citizens, updated as context shifts
Population indicators are distinguished from performance indicators	Not distinguished; one blended set of numbers used for both purposes	Distinction understood by some staff, not applied consistently	A defined, held population indicator set alongside program performance measures	Population trends analyzed for cause before responsibility is assigned	Population indicators co-owned across agencies and partners, jointly resourced

Checklist

Draw a logic model (inputs, activities, outputs, outcomes, impact) for each major program before finalizing its indicators.
Write the theory of change's causal assumptions specifically enough that evidence could contradict them.
Separate effort, quality, and effect measures in every performance report, even outside a formal results-based accountability implementation.
Build or update the Public Value Scorecard so all three legs — legitimacy, value, capacity — are reported together, not separately.
Choose a small, stable set of population-level headline indicators and commit to holding them across several reporting cycles.
Check current reports for language that describes an output using outcome words, and correct it.
Publish the outcomes framework ahead of the next budget or funding round, not as its aftermath.
Name explicitly what the organization does not solely control for any population-level claim it reports.
Set a fixed cycle for revisiting logic models against actual evidence, not only at a scheduled formal evaluation.
Identify at least one external stakeholder to review the logic model's assumptions before it is finalized.
Audit the current indicator set for opportunities to retire measures nobody uses in an actual decision.
Agree, in advance, what the organization will do differently if a headline outcome fails to move for two reporting cycles.

Key sources

Mark H. Moore, Creating Public Value: Strategic Management in Government (1995) and Recognizing Public Value (2013) — the originating texts for the Public Value Scorecard, adapting the balanced scorecard tradition to the three legs of the strategic triangle.
Mark Friedman, Trying Hard Is Not Good Enough: How to Produce Measurable Improvements for Customers and Communities — the originating text for results-based accountability, including the population-accountability/performance-accountability distinction and "turn the curve" analysis.
Robert Kaplan and David Norton's balanced scorecard literature — the private-sector source Moore adapted for the public value context.
OECD guidance on results-based management and evaluation terminology, used across development and public-administration practice to standardize outcome and impact language.
World Bank Independent Evaluation Group guidance on results frameworks and theories of change for development operations.
Vermont Agency of Human Services' public account of applying results-based accountability at state and community level — one example of the framework in live government use.

References

Mark H. Moore — Wikipedia — https://en.wikipedia.org/wiki/Mark_H._Moore
Public value — Wikipedia — https://en.wikipedia.org/wiki/Public_value
Balanced scorecard — Wikipedia — https://en.wikipedia.org/wiki/Balanced_scorecard
Logic model — Wikipedia — https://en.wikipedia.org/wiki/Logic_model
Theory of change — Wikipedia — https://en.wikipedia.org/wiki/Theory_of_change
Monitoring and evaluation — Wikipedia — https://en.wikipedia.org/wiki/Monitoring_and_evaluation
Program evaluation — Wikipedia — https://en.wikipedia.org/wiki/Program_evaluation
Performance indicator — Wikipedia — https://en.wikipedia.org/wiki/Performance_indicator
Goodhart's law — Wikipedia — https://en.wikipedia.org/wiki/Goodhart%27s_law
Mark H. Moore, faculty profile — Harvard Kennedy School — https://www.hks.harvard.edu/faculty/mark-moore
Results-Based Accountability — Vermont Agency of Human Services — https://humanservices.vermont.gov/our-impact/results-based-accountability
Glossary of Key Terms in Evaluation and Results-Based Management for Sustainable Development (Second Edition) — OECD — https://www.oecd.org/content/dam/oecd/en/publications/reports/2023/06/glossary-of-key-terms-in-evaluation-and-results-based-management-for-sustainable-development-second-edition_2767e14e/632da462-en-fr-es.pdf
Designing a Results Framework for Achieving Results: A How-to Guide — World Bank Independent Evaluation Group — https://documents1.worldbank.org/curated/en/331541563854787772/pdf/Designing-a-Results-Framework-for-Achieving-Results-A-How-to-Guide.pdf