Shared savings is often presented as a clean solution: spend less downstream, reinvest upstream. In practice, shared savings within integrated funding pilots is difficult because savings can be “created” by restricting access, shifting cost to another payer, or changing coding rather than changing outcomes. Funders back these pilots when the operating rules make gaming hard and quality failure visible.
This article supports Integrated Funding Pilots and is designed to pair with measurement approaches in Using Data for Commissioning & Oversight.
Why shared savings needs quality gates in community-based systems
Community services operate close to risk: safeguarding, medication safety, crisis escalation, and housing stability. If a pilot pays for lower utilization without checking whether people are stable, the model can unintentionally reward harm. Quality gates are the mechanism that ensures savings are only “real” when outcomes and safety indicators hold.
Most credible pilots define: (1) a baseline and attribution method, (2) a savings calculation approach, (3) a quality gate framework, (4) reinvestment rules, and (5) escalation and remediation processes when performance drifts.
Operational Example 1: Quality-gated shared savings for avoidable acute use
What happens in day-to-day delivery
A pilot defines an attributed population and tracks avoidable ED use and short-stay admissions monthly. A performance team produces a standardized report showing utilization, follow-up timeliness, medication continuity indicators, complaint and incident trends, and equity breakdowns. If utilization improves and quality gates pass, a portion of savings is released into a reinvestment pool. Operational leaders agree reinvestment priorities (extra care management capacity, after-hours coverage, transportation supports) and document decisions with an implementation plan and milestones.
Why the practice exists (failure mode it addresses)
This practice prevents the failure mode where utilization drops because access was constrained or individuals disengaged, rather than because community support improved.
What goes wrong if it is absent
Without quality gates, organizations may reduce services to “create” savings, delay referrals, or avoid high-risk individuals. Harm presents later as crisis escalation, safeguarding concerns, poorer experience, and inequitable outcomes that undermine the pilot.
What observable outcome it produces
A gated model produces defensible improvement: reduced avoidable acute use alongside stable or improved safety, engagement, and equity indicators. Evidence includes consistent reporting, documented reinvestment decisions, and clear links between reinvestment and subsequent outcome trends.
Operational Example 2: Attribution and denominator control for multi-agency pilots
What happens in day-to-day delivery
The pilot establishes rules for who counts in the attributed population (enrollment criteria, minimum engagement, timing rules) and locks the denominator for each measurement period. A data governance group validates rosters, resolves disputes, and publishes a change log when membership shifts (moves, deaths, eligibility changes). Providers receive a roster with identifiers, risk stratification, and “known exclusions” to prevent informal manipulation. Monthly reconciliation checks for unusual churn, late exclusions, or patterns that suggest denominator gaming.
Why the practice exists (failure mode it addresses)
Shared savings becomes meaningless when attribution is unstable. If the denominator can be changed after the fact, results can be engineered by removing high-cost individuals rather than improving care.
What goes wrong if it is absent
Partners argue about who was “in scope,” performance becomes contested, and trust breaks down. Providers may avoid complex cases, and funders lose confidence that reported savings reflect real system change.
What observable outcome it produces
Stable attribution produces credible reporting, fewer disputes, and a clear link between service model changes and outcomes. Audit evidence includes roster lock dates, change logs, dispute resolutions, and documented application of rules.
Operational Example 3: Reinvestment governance that prevents drift and ensures equity
What happens in day-to-day delivery
A reinvestment committee (funder, provider, community representation) approves how savings are used. Requests must show: the operational change being funded, the failure mode it addresses, the expected measurable outcome, and how it will be monitored. Funds are released in stages tied to delivery milestones (staff hired, protocols implemented, training completed, dashboards live). Quarterly reviews assess whether reinvestment produced the intended effect, with the option to stop or redirect funding if outcomes do not improve.
Why the practice exists (failure mode it addresses)
This practice prevents reinvestment from becoming discretionary spending that cannot be defended, and it protects against widening inequity by ensuring dollars target gaps revealed in data.
What goes wrong if it is absent
Savings get absorbed into general budgets, partners disagree on priorities, and the pilot loses legitimacy. In worst cases, reinvestment funds changes that are popular but not effective, and outcomes stagnate.
What observable outcome it produces
Effective reinvestment governance produces visible capacity increases tied to measurable improvements (timeliness, engagement, reduced duplication, improved equity indicators). Evidence includes committee minutes, staged release documentation, milestone tracking, and pre/post impact assessment.
What funders explicitly require before they will scale a shared savings pilot
Expectation 1: A credible savings methodology and anti-gaming controls. Funders expect transparent baselines, clear attribution rules, risk adjustment where appropriate, and defined exclusions. They also expect routine checks for denominator manipulation, coding shifts, and sudden access changes.
Expectation 2: Safety, quality, and equity gates with escalation authority. Oversight bodies expect explicit thresholds that can pause or reduce shared savings if complaints, incidents, safeguarding concerns, or inequity indicators worsen. They expect a remediation process with timelines, owners, and documented follow-up.
Common design mistakes and how to avoid them
Pilots fail when they (1) rely on one metric, (2) define savings without showing stability, (3) leave reinvestment vague, or (4) lack a mechanism to intervene when performance drifts. Strong pilots build a disciplined operating cadence: monthly performance reviews, quarterly deep dives, a clear exception process, and independent validation of key indicators.
Shared savings can fund real prevention and coordination. But it only works when the rules make it easier to do the right thing than to game the system—and when governance makes problems visible early enough to fix them.