When activity replaces outcomes

A service can answer more calls, process more forms and close more cases while helping fewer people. If its performance system treats those counts, speed and cost per transaction as sufficient evidence of success, deteriorating outcomes can coexist with an improving performance report.

The core thesis is that efficacy and participant outcomes can be displaced by the pursuit of apparent efficiency and the quantity of interactions. When institutions are rewarded for processing people without establishing whether they have helped them, recorded activity can replace achieved purpose.

This article examines Nathan Organ’s framing of designed ineptitude through that measurement problem. Administrative friction, repeated contacts, displaced executive work and complaint containment are possible mechanisms through which a service can retain the appearance of productivity while failing its purpose. Intent remains a separate question: the pattern can emerge, be knowingly tolerated, or be deliberately maintained.

Research on administrative burden, administrative exclusion and sludge establishes how procedures can obstruct access without a formal refusal. The present analysis connects those mechanisms to a prior question: what is the institution actually being measured and rewarded for achieving? [1–3]

The central question is whether an organisation can report improving performance while the people it exists to serve receive less benefit, and what measurement and accountability arrangements allow that divergence to persist.

Scope and approach

This is a critical narrative review and systems analysis, prepared on 4 November 2026. It draws on peer-reviewed research, Australian audit findings, official investigations and complaint-management guidance. International experiments help test causal mechanisms; Australian research and oversight reports establish their relevance to local service systems.

It is not a systematic review, a new participant study or a prevalence estimate. The sources vary in strength: experiments can establish effects within particular settings; interviews illuminate experiences; administrative data reveal patterns; audits assess defined periods and functions. Neither an association nor a poor performance measure establishes intent by itself. Historical findings are identified by their study or audit period rather than presented as a live assessment of every service.

The core measurement failure: activity replaces outcomes

Four questions need to remain distinct. How much work was recorded? What resources did it consume? Did it achieve the intended benefit? Who received that benefit, and who could not obtain it?

The Department of Finance distinguishes output measures from efficiency and effectiveness. Outputs describe what is produced; efficiency relates resources to outputs while retaining attention to quality; effectiveness concerns achieving the intended result. The department acknowledges that output measures have historically dominated performance reporting. [21]

Here, efficacy means whether the service delivers its intended real-world benefit; effectiveness is the terminology generally used in the public-service evaluation sources. A referral issued is an output. Whether it leads to appropriate assistance is a further question. A complaint closed is an administrative event. Whether a justified remedy was delivered and worked requires separate evidence.

Counting interactions does not establish efficiency without a resource measure, and cheaper interactions do not establish effectiveness. The Productivity Commission distinguishes technical efficiency from cost-effectiveness and warns that improving access for disadvantaged groups can raise average unit costs. A more expensive service can still deliver better value through better outcomes. [23]

The proposed failure is therefore specific: organisations can mistake easily counted activity and narrow cost measures for evidence that their purpose is being fulfilled. Useful efficiency measures belong alongside verified outcomes, accessibility and quality. They cannot stand in for all three.

Australian oversight identifies the same gap

The ANAO’s report published on 5 February 2026, covering 21 entities’ 2024–25 performance statements, found that many agencies primarily reported activities rather than achievements. Outcomes were less frequently measured; even measures labelled effectiveness often missed longer-term results. It also warned that narrow efficiency targets can favour volume over quality and discourage serving high-need people. [22]

An important qualification strengthens the diagnosis: the ANAO found genuine efficiency measures were rare too. Timeliness measures alone did not establish efficient resource use. The evidence supports a failure to connect resources, activity and benefit, rather than a blanket claim that governments rigorously measure efficiency while never assessing outcomes. [22]

How the figures can improve while the service deteriorates

Consider a hypothetical service with the same eligible population, comparable needs, the same follow-up period and an operating cost of $100,000. In its first period, it records 1,000 interactions and achieves 500 independently verified resolutions. Its cost is $100 per interaction and $200 per resolution.

In the second period, it records 2,000 interactions but achieves only 250 comparable resolutions at the same total cost. Reported interaction volume has doubled and cost per interaction has halved to $50. Yet the number of achieved outcomes has halved and cost per resolution has doubled to $400. These are illustrative figures, not findings about a particular organisation.

The same performance change looks favourable when measured per contact and unfavourable when measured per achieved outcome. A transaction-based report can therefore celebrate the wrong improvement. If contacts have also become less useful, lower unit cost would not establish a quality-adjusted efficiency gain.

Two different routes can conceal failure. Someone repeatedly calling about one unresolved problem generates additional recorded interactions. Someone exhausted by the process eventually stops calling and disappears from the active workload. In the proposed model, continued struggle can be counted as engagement while abandonment can be interpreted as reduced demand. Neither establishes that the underlying need was met.

Repeated contact is not inherently waste: ongoing support, trusted relationships and complex work may require it. The relevant distinction is between contact that contributes to an agreed purpose and repetition caused by avoidable failure. Likewise, reduced contact can reflect successful assistance or exclusion. Counts alone cannot tell those possibilities apart.

The most disturbing possibility is that an organisation can improve its apparent efficiency by reducing the number of people who successfully access its services. This depends on what is counted and whose costs are included. Fewer users do not automatically improve a genuine efficiency ratio; fixed costs may make unit costs worse. The concern is that exclusion can make workload, speed or expenditure indicators look favourable while unmet need is omitted.

The mechanisms already have names

Moynihan, Herd and Harvey’s influential administrative-burden framework distinguishes learning costs, such as discovering eligibility and requirements; compliance costs, such as completing forms and supplying evidence; and psychological costs, including stress, stigma and diminished autonomy. Their analysis also treats the distribution of these burdens between government and individuals as a policy choice, sometimes made without the visibility of an explicit eligibility change. [1]

Brodkin and Majmundar describe administrative exclusion: non-participation attributable to organisational processes rather than a person’s preferences or formal eligibility. This directs attention to people who never reach an assessable application or an appealable decision. [2]

The OECD describes sludge as unjustified friction that obstructs access and increases psychological burden. Its 2024 report sets out a structured approach to identifying and reducing these obstacles, drawing on ten international case studies. Friction is therefore something that can be investigated and redesigned, rather than accepted as an unavoidable feature of administration. [3]

Designed ineptitude is used here as an integrating research question, not a claim to have discovered these mechanisms. Their interaction becomes especially concerning when output-based reporting fails to reveal whether the institution fulfils its purpose.

Friction changes who receives assistance

The strongest evidence goes beyond reports of frustration.

Deshpande and Li’s 2019 study examined US Social Security Administration field-office closures. In affected areas, disability applications fell by about 10%, while the number of new disability recipients fell by about 16%. Effects were particularly substantial for applicants with moderately severe conditions and lower education. The published study concluded that closures reduced targeting efficiency under existing eligibility standards. Increased congestion at neighbouring offices appeared more important than additional travel or information costs. [4]

This challenges the assumption that difficult applications reliably screen out people who do not need help. In this setting, reduced assistance also screened out people who would otherwise have qualified.

A randomised experiment involving approximately 30,000 older people likely eligible for the US Supplemental Nutrition Assistance Program provides complementary evidence. Over nine months, enrolment was 6% in the control group, 11% among people given eligibility information, and 18% among those offered information plus application assistance. Practical assistance achieved considerably more than information alone. [5]

There is an important qualification: those induced to enrol by the interventions were generally healthier and financially better off than control-group enrollees. Increasing uptake did not automatically prioritise the most disadvantaged people. The authors interpreted the findings as consistent with greater barriers among those with greater need. [5]

The implication for service design is specific. Publishing clearer information can help, but a person may also need assistance with completing the process. Even then, aggregate improvement should be checked for unequal benefits across groups.

Australian disability research shows the same problem

Carey, Malbon and Blackwell’s 2021 systematic review found substantial administrative complexity in the NDIS, with burdens experienced unevenly across social groups. Their analysis identifies a tension within personalised funding: increasing individual control can also increase the amount of knowledge, coordination and advocacy required to exercise it. [6]

Yates and colleagues subsequently interviewed 30 women with disability who were NDIS participants or had considered applying. Their 2022 study documented significant administrative burdens that impeded access to sufficient support. These interviews provide evidence about mechanisms and lived experience; they do not establish how frequently every NDIS participant encounters them. [7]

A 2025 study by Disney and colleagues analysed NDIS administrative data from 2016–22. Among 485,676 applicants with recorded decisions, eligibility inequalities were particularly evident for certain disability groups, including physical and psychosocial disability. Differences were not uniform: some groups had high approval rates and comparatively small demographic inequalities. The study does not isolate administrative burden as the cause of every disparity. [8]

Its exclusions also expose a measurement problem. The eligibility analysis excluded 136,199 applications classified as incomplete, cancelled or withdrawn, about 19% of the original applicant cohort. Those records cannot all be attributed to obstructive administration. Nevertheless, an analysis of completed decisions cannot, by itself, explain what happened to people who did not reach that stage. [8]

A credible access assessment therefore needs two questions: whether decisions are fair, and whether people can reach a decision at all.

The displacement of executive work

The role of executive functioning is already recognised in this literature. Christensen and colleagues’ 2020 conceptual analysis describes a catch-22: circumstances that increase a person’s need for assistance can also reduce the cognitive resources available to navigate its administrative requirements. Their model considers planning, initiation, remembering, monitoring and managing responses to stressful interactions. It is a theoretical synthesis with illustrative evidence, not a direct experimental test of every proposed pathway. [9]

For this article, executive-burden displacement describes a practical organisational pattern: the service makes the recipient responsible for integrating its fragmented processes.

Consider a hypothetical applicant who must discover the correct team, obtain records already held elsewhere, explain contradictory instructions, track several deadlines and repeatedly ask whether anyone has acted. The institution has divided its work between departments. The applicant becomes the only person responsible for making those departments function as a service.

This is an allocation-of-work problem. It should not be reduced to a judgement about the applicant’s intelligence, effort or character. The relevant questions are what work is required, who can reasonably perform it, and what assistance is available.

Burden can also affect wellbeing. Across three Danish studies, Baekgaard and colleagues found that reduced compliance demands were associated with greater perceived autonomy; two studies also found reduced stress. The designs included a survey experiment and field studies around the sudden suspension of requirements during the COVID-19 lockdown. These findings support treating administrative demands as a potential contributor to distress, rather than assuming distress comes only from the circumstances that led someone to seek help. [10]

The proposed feedback loop is straightforward: a difficult process consumes capacity; reduced capacity makes completion harder; incomplete participation produces additional demands or closure. Whether that loop occurs, and for whom, must be measured rather than presumed.

How poor access can become useful to an institution

The following systems hypothesis connects the measurement problem to organisational incentives; it is not established for every provider.

Suppose an organisation is rewarded mainly for remaining within budget, answering contacts or closing cases. Suppose it has weaker obligations to identify unmet need, follow people who abandon applications, or verify that referrals resulted in assistance. Under those conditions, reducing the number of people who progress through the system may relieve operational pressure without appearing as a failure in its main performance measures.

A plausible cycle follows. Public communications generate expectations. Administrative demands filter who continues. People with fewer resources leave earlier. Reporting describes those who remain. Management sees manageable workloads or completed transactions, while the costs of exclusion are borne by individuals, families and other services.

No coordinated conspiracy is necessary for this cycle. Separate teams can follow locally sensible rules while nobody owns the person’s complete journey. Nor does the cycle require idle employees: heavily pressured frontline staff can be caught inside it.

Financial incentives must be checked rather than assumed. Under a hypothetical fixed operating budget, less demand may reduce immediate workload. A provider paid only for delivered services may lose revenue when people leave, while still finding complex cases financially difficult. A grant, contract or performance regime may reward activity without adequately measuring benefit. These arrangements create different incentives; they cannot all be described as profit from non-delivery.

The critical research task is to trace the actual funding conditions, management measures and consequences of failure. Who gains capacity or avoids scrutiny when an applicant disappears? Who bears the resulting cost? Who is required to notice?

A service can become efficient at processing transactions while remaining ineffective at accomplishing its purpose. Calling the resulting exclusion “non-engagement” does not explain its cause.

Complaints can be processed without producing correction

Complaints are supposed to provide a route from an adverse experience to a response and, where warranted, organisational change. That route can fail at intake, recognition, escalation, investigation, remedy or verification.

The NSW Ombudsman explicitly cautions that few complaints do not necessarily indicate good performance. Its guidance identifies power imbalances, fear of detrimental consequences, complex procedures, lack of support and the belief that complaining is pointless as barriers. It also sets out accessible channels, ownership and escalation as features of effective complaint management. [13]

This suggests a test for what this article calls complaint containment: can a complaint be acknowledged, counted and closed without the underlying issue reaching someone able to act? The term describes a possible functional outcome. It does not establish that a call-centre employee intended to suppress the complaint.

Australian oversight findings demonstrate why the question is reasonable. The ANAO’s January 2026 Age Pension audit examined the older Australians phone line over 2021–22 to 2023–24. It found that 34.61% of recorded calls were terminated by Services Australia as congested. Callers could not remain in the queue, and the agency lacked data on whether they subsequently used the suggested self-service option. These were call attempts, not counts of distinct people denied assistance. [11]

The same audit found that monthly complaint summaries were supplied to senior executives, but Services Australia could not provide the ANAO with an account of actions executives had taken in response. That establishes a gap in demonstrated follow-through, rather than proving that no action occurred. [11]

Separately, the ANAO’s September 2025 audit of the NDIS Quality and Safeguards Commission found that its 2021 complaints manual had ceased being used in 2023 without replacement. It reported off-system complaint records and locally adapted processes associated with operating-system usability problems. It also identified deficiencies in quality assurance and inconsistencies between performance-reporting datasets. [12]

Those findings concern the NDIS Commission, which regulates providers, rather than the NDIA, which administers participant access and plans. Conflating them would obscure where responsibility actually sits.

For evaluation, complaint closure should be separated from a reasoned decision, a completed remedy and confirmation that the remedy worked. A justified rejection may properly close a complaint. A referral may be appropriate. Neither should automatically be counted as delivery of the assistance originally sought.

When intent can be demonstrated

Three evidentiary categories should remain distinct.

Emergent obstruction arises from fragmented responsibilities, poor systems, inadequate capacity or conflicting requirements. Establishing it requires evidence of the process and its effects, not evidence of malicious intent.

Knowingly tolerated obstruction involves decision-makers being credibly informed of avoidable barriers yet leaving them in place without adequate justification or mitigation. Evidence should address what was known, who could act, what options existed and what response followed. This is an analytical category, not a legal finding of negligence.

Deliberate obstruction or concealment requires stronger evidence: instructions, communications, decisions or investigative findings showing an intention to deter access, misrepresent performance or prevent effective scrutiny. A bad outcome alone is insufficient.

There are documented Australian examples of intentional obstruction of accountability. In its March 2026 Robodebt investigation, the National Anti-Corruption Commission found serious corrupt conduct by two of the six referred individuals, including intentionally misleading the Commonwealth Ombudsman during its 2017 investigation. It found that the other four had not engaged in corrupt conduct. The Commission also found insufficient admissible evidence to establish alleged criminal offences beyond reasonable doubt and did not refer the two individuals for prosecution. These are corruption findings, not criminal convictions. [14]

That case demonstrates that deliberate interference with scrutiny can occur. It does not establish that ordinary queues, application requirements or all complaints processes were designed for that purpose.

Responsibility also need not wait for proof of an original plan. Once a harmful barrier is documented, an institution’s subsequent decisions become a separate object of scrutiny.

“Free” does not mean the recipient owes submission

The absence of a charge at the point of use says little about the recipient’s actual costs. They may still contribute time, transport, documentation, lost income and extensive coordination. An institution’s operating expenditure and a recipient’s unpaid work should be considered together when evaluating efficiency.

The ethical problem becomes sharper when assistance is treated as a personal favour, with gratitude expected to replace scrutiny. Appreciation for an individual worker can coexist with a legitimate criticism of the service. Gratitude is not a reliable measure of adequacy.

Australian research supports that distinction. Booth and colleagues interviewed 14 adults using two charitable food services in inner-city Perth. Their 2018 study documented appreciation alongside resignation, inadequate choice and damage to dignity. Participants described substantial effort involved in obtaining food and proposed practical improvements. The study is small and setting-specific, but it demonstrates why a grateful response cannot be assumed to mean a service meets people’s needs. [15]

Deservingness judgements can also shape acceptance of burdens. A study of 495 Danish local politicians found that willingness to impose administrative demands varied with perceived claimant deservingness, political orientation and personal experience of receiving benefits. Information about claimants’ psychological costs did not itself reduce tolerance of those burdens. This concerns policymaker attitudes in a particular setting, not proof of a universal attitude among Australian service workers. [16]

The normative conclusion is that respectful access should not depend on performing deference. Public eligibility rules can be explained and applied; charitable services can honestly state their capacity. Neither requires treating the recipient’s dignity as discretionary.

The Robodebt Royal Commission’s recommendations provide an institutional counterpart: design around recipients, avoid stigmatising language, provide suitable communication options and take practical steps to avoid exacerbating distress. These were recommendations for reform, not evidence that every recommended practice has been implemented in every service. [17]

The evidence also limits the argument

A serious analysis must retain evidence that resists its initial framing.

The ANAO’s March 2025 audit rated the ATO’s complaint management largely effective. It found accessible channels and allocation to appropriately experienced resolvers, while recommending better documentation, root-cause analysis and public reporting. This is incompatible with a blanket claim that government complaint systems exist only to absorb criticism. [18]

Nor is public ownership itself an adequate explanation. In a Flemish nursing-home field experiment, Jilke and colleagues found no significant overall difference in response rates to enquiries using Flemish or Maghrebian names. However, private facilities were about 20 percentage points less likely to provide enrolment information to the Maghrebian-name enquiry; the same disparity was not found in public facilities. Counting replies alone would have missed an important difference in usable assistance. [19]

That experiment supports examining the content and usefulness of a response. It also cautions against assuming that private or community provision automatically fixes public-sector problems.

Some friction serves legitimate purposes. Identity checks, consent, evidence and review can protect people and public resources. NSW’s sludge-audit guidance expressly distinguishes justified safeguards from excessive or unjustified steps. The task is to assess proportionality and less burdensome alternatives, not remove every requirement. [20]

The reviewed evidence therefore supports administrative exclusion as a real phenomenon. It does not quantify how prevalent intentional obstruction is across government and community services, establish that those sectors are uniquely affected, or justify allegations about individual staff members’ motives.

A testable research framework for NableU

NableU could investigate the central hypothesis through a participant-led study of complete service journeys: do improvements in throughput, handling time or cost per interaction coincide with improved participant outcomes, or can those trends diverge? The study would test specified mechanisms rather than begin by classifying organisations as deliberately inept.

Define intended outcomes with participants before evaluating success. Assistance received, an agreed problem resolved, maintained wellbeing and prevented deterioration can all be relevant. Contact volume and cost should be linked to those outcomes, with follow-up suited to the service. Report unknown outcomes and abandonment separately, preserve information about complexity and unmet need, and distinguish the service’s contribution from external influences. An outcome framework must not reward selecting easy cases or imposing institutional goals on participants.

Follow people from initial need, not only completed applications. Recruitment should include people who considered applying, abandoned the process, were rejected, obtained help and remain waiting. A sample drawn only from successful users would miss part of the problem; a sample drawn only from complainants would create a different bias. Recruitment routes and limits should be reported.

Measure the work and where it is transferred. A journey record could capture active time, elapsed waiting, repeated information requests, hand-offs, inaccessible contact requirements, assistance received and recovery time reported by participants. Different burdens should remain visible rather than being prematurely collapsed into one numerical score. Documentation should be optional and low-effort; research about administrative burden must not impose another ordeal.

Distinguish outcomes carefully. Record eligibility decisions, support actually received, unresolved need, withdrawal and unknown outcomes separately. Failure to reach someone does not establish that their need has disappeared. Abandonment rates require a defined starting cohort and follow-up period; unresolved cases still in progress must not silently become either successes or failures.

Trace accountability and incentives. Examine complaint recognition, access to an authorised resolver, reasons for closure, completed remedies and changes to recurring failures. Where records are available, compare these with contract requirements, budgets, performance measures and documented warnings. Record competing explanations, including staffing shortages, lawful eligibility restrictions and genuinely unavoidable dependencies.

Useful tests would compare standard pathways with assisted navigation, fewer duplicated requests or a named case owner. A staggered implementation or an appropriate comparison group would be stronger than relying on a simple before-and-after change. Existing support should not be withheld to create an experiment. Where formal human research is undertaken, obtain the appropriate ethics review and accessible consent.

The hypothesis would be weakened where records show low unjustified burden, effective remedies, verified follow-through and no exclusion associated with the suspected barriers. It would be strengthened where particular barriers consistently predict attrition, reducing them improves access, and records show that decision-makers knew about the problem but declined feasible correction.

Evidence that a barrier has an exclusionary effect still would not, by itself, establish intent. That additional conclusion needs its own evidence.

A proposed Institutional Friction Index could eventually assist comparison, but it should begin as a transparent set of measures. Weights, reliability, validity and sensitivity to different support needs would need testing before publishing league tables or presenting a score as an objective verdict.

Design obligations that follow

The practical aim is to reduce the work required to exercise agency while preserving the person’s control over decisions.

Services should offer a reachable person or team with clear responsibility for the complete issue, continuity across hand-offs, usable written information and communication options suited to the recipient. Records should be reusable where lawful and authorised. A person should not have to repeatedly reconstruct a case because internal systems do not communicate.

Performance reporting should link resources and activities to participant-valued results, while distinguishing a contact answered, an application decided, a referral made, assistance received and an unresolved need. Cost per interaction should be accompanied by evidence of service quality, outcomes and who remains excluded; lower cost alone is insufficient. Complaint systems should provide reasons, review routes and a traceable record of corrective action. Staff need the authority, training and capacity to act; simply demanding warmer language from an under-resourced call centre will not repair missing decision-making pathways.

These are proposed design requirements informed by the reviewed evidence, not a claim that one interface can solve every funding or policy constraint. Their cost should be assessed alongside recipient labour, repeated contacts and downstream consequences. Moving an administrative task off the organisation’s balance sheet does not make it disappear.

NableU should apply the same scrutiny to its own systems. Participant control includes deciding what is shared, choosing assistance, declining unnecessary demands and seeing what happens next. A participant-first purpose is credible only where those choices remain usable through the actual process.

Conclusion

The organising problem is the substitution of recorded activity and apparent efficiency for evidence that a service achieves its intended purpose. A system can process more interactions, reduce reported unit costs and close more cases without producing better outcomes. Whether that occurs in a particular service requires linked evidence about resources, access, quality and results.

Administrative-burden research establishes mechanisms through which access can fail, while Australian oversight identifies gaps between activity reporting and outcome assessment. Those findings support investigating how institutions can appear successful while people remain unsupported; they do not establish universal deliberate obstruction. [1–8,21–23]

The ethical standard should not stop at whether an organisation intended the original failure. Once it knows that its measures reward throughput while concealing unmet need, its response becomes part of the evidence.

A service should demonstrate what its interactions accomplish for people. Efficiency should describe the responsible use of resources to fulfil that purpose, with outcomes, quality and access kept visible. An institution that cannot establish whether people are helped has not established that more activity represents better performance.

Sources and further reading

This critical narrative review draws on 23 academic and official sources. Experimental, observational, qualitative and audit findings answer different questions; none alone establishes widespread intent. References support the numbered citations in the article. Prepared 4 November 2026.

NableU Community Systems Incorporated advocates for dignity, agency, autonomy, cognitive equity and cognitive sovereignty. About NableU · Back to the article title