ATO's real-time data models find unpaid super in 90% of cases

The ANAO reports that the ATO's real-time superannuation guarantee models lifted the average share of compliance cases that raise a liability from about 60–70% to 90%.

  • 2 for an organization

    The ANAO reports that the share of SG compliance cases raising a liability rose from about 60–70 percent to 90 percent with the new models.

  • Not rated a person

    Employees gain when unpaid super is raised as a liability, but the report gives no amounts recovered through these models.

Depth runs from 1, a convenience or time saved, to 5, a life or a livelihood preserved.

Status
Partly reported benefit.
Evidence

foundstudiedreportedclaimed

An independent descriptive report supports the benefit.

Compared with

The ATO's earlier SG case selection with the High Risk Employer and Stratified Sampling Nearest Neighbour models. Measured.

Without the system
approximately 60–70 per cent
With it
90 per cent
The system’s part
Not isolated. The report says that one model compares employers with similar groups of employees and that the other uses payroll and fund data to identify under-reporting. It does not say that either learns from data, and it says that both rely on Excel work by a few staff. The strike rates compare the new models with the old ones, and the report does not say how they were measured. Elsewhere it notes that high strike rates in SG Taskforce audits may partly reflect the choice of high-risk employers.
Not yet checked
LimitsCosts and harmsLearns from dataLater outcome

Download data Cite Suggest a correction

Benefit type
Improved Task Accuracy
System
ATO superannuation guarantee real-time data models (Employer Simulation and Evaluation Model and SG Under Reporting model)
Made by
Australian Taxation Office (the report names no other developer)
Used by
Australian Taxation Office (ATO)
Where
Australia (national)
Stage
In a pilot
Date
, when the source was published
Kind
Aggregate result
Broad benefit
Evaluating Information to Determine Compliance with Standards, Processing Information
Last revised
1 November 2026

Sources

  1. Descriptive report, by authors with no tie to the system.

    Australian National Audit Office. Addressing Superannuation Guarantee Non-Compliance. Auditor-General Report No. 24 of 2021–22, published 28 April 2022. Chapter 2, paras 2.45-2.55 (report web page, Internet Archive copy).

    The new models increased the average strike rate from approximately 60–70 per cent to 90 per cent. However, the new models require a high level of data validation, involve limited sampled data sets and compliance cases, are manually intensive, and require intensive use of Microsoft Excel by a few experienced staff.

Details

AI system

Workflow

Employers reported pay through Single Touch Payroll and super funds reported contributions through the Member Account Transaction Service. The Employer Simulation and Evaluation Model compared each employer with similar groups of employees, and the SG Under Reporting model used payroll and fund data to identify under-reporting. Running the models took heavy data validation, manual work and intensive use of Microsoft Excel by a few experienced staff.

Follow-up

Conditions for replication

Payroll and fund data must be matched reliably, and the ATO estimated that data integrity issues affected 10 to 15 percent of the data. Enough experienced staff must be available, since the models are manually intensive and use limited samples.

Entry details

Sector

Tax administration: superannuation guarantee compliance.

Purpose

The Australian Taxation Office built the real-time data models to select employers who are likely to have underpaid the superannuation guarantee for compliance cases.

31 more questions not yet checked

model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.

Cite this record

Peter H. Vartanian. (2026). ATO's real-time data models find unpaid super in 90% of cases (AIBD-1039). Artificial Intelligence Benefits Database, edition of 1 November 2026. https://petervartanian.xyz/aibd/records/AIBD-1039.html (archived at https://doi.org/10.7910/DVN/AADSII, version 7.0)

BibTeX Data Field definitions