Success against femicide in South Africa cannot be measured by the number of speeches, campaigns or new commitments announced. A credible test would show whether women at risk are identified and protected earlier, whether reports receive timely and competent responses, whether protection orders are enforced, whether investigations reach court, whether prosecutions result in accountability, and whether survivors can actually access safe accommodation, healthcare, counselling and legal support. South Africa’s National Strategic Plan on Gender-Based Violence and Femicide (NSP-GBVF) already calls for measurable, time-bound targets and monitoring across government and society.
The hardest measure is also the most important: whether fewer women are killed. Everything else is an intermediate indicator. Arrests, convictions, protection orders and service access can show whether institutions are functioning, but none of them by itself proves that prevention is working. A useful national scorecard therefore needs to connect the chain from risk and reporting to police response, protection, investigation, prosecution, survivor support and ultimately changes in femicide and related violence.
The first question: are fewer women being killed?
The clearest outcome measure is the number and rate of femicides over time.
But this is more complicated than simply counting murders of women.
South Africa’s crime statistics system has experienced data-quality problems. In 2025, SAPS withdrew previously published disaggregated statistics on crimes against women and children after identifying discrepancies, duplicate counts and problems with linking victims to charges. The police said the affected statistics would be reinstated after correction and validation.
That matters because a country cannot reliably evaluate progress without reliable measurement.
A serious femicide monitoring system therefore needs to establish:
- How many women and girls were killed
- Their age and location
- Their relationship to the alleged perpetrator
- Whether previous GBV was reported
- Whether a protection order existed
- Whether police or other authorities had previously received warnings
- Whether a firearm or other weapon was involved
- Whether the case is still under investigation, before court or finalised
The objective is not merely to produce a national number. It is to identify patterns that can inform prevention.
Reporting should become easier—and safer
An increase in reports is not necessarily evidence that violence has become worse.
It can also mean that victims are more willing and able to report.
That makes reporting rates an important but complicated indicator.
A functioning system should track not only the number of reports, but what happens after someone reports.
Useful measures include:
| Indicator | What it tells us | What success would look like |
|---|---|---|
| GBV reports received | Whether cases are entering the system | Accessible reporting with fewer barriers |
| Response time | How quickly authorities act after a report | Faster responses, particularly for high-risk cases |
| Risk assessment | Whether danger is identified early | Consistent assessment of escalation and repeat violence |
| Protection orders | Whether legal protection is being sought and granted | Accessible orders with timely decisions |
| Protection-order breaches | Whether orders are being respected and enforced | Rapid intervention after breaches |
| Investigations | Whether reported cases are actively pursued | Fewer unresolved or poorly investigated cases |
| Prosecutions | Whether viable cases reach court | Stronger case progression and fewer avoidable failures |
| Convictions | Whether cases ending in court produce accountability | Reliable, transparent conviction data |
| Survivor services | Whether victims receive practical support | Timely, equitable access to healthcare, shelter, counselling and legal assistance |
| Femicide rate | Whether the ultimate harm is declining | Sustained reduction over time |
No single measure should be interpreted in isolation.
Response time matters most when risk is escalating
For someone facing immediate danger, the difference between a prompt response and a delayed response can be consequential.
South Africa’s current policy discussions recognise the importance of institutional response. In 2026, the police department said it was examining the administration of domestic violence at police stations, the availability of private interviewing facilities and complaints involving poor responses to GBV cases.
That points toward a more useful metric than simply counting police interventions.
Authorities should be able to measure the time between:
report → risk assessment → police response → protection measure → investigation.
The relevant target will vary by the seriousness and circumstances of a case. An emergency call involving an immediate threat should not be treated the same way as a historical report with no indication of current danger.
The point is to make response standards measurable rather than rhetorical.
Protection orders should be measured from application to enforcement
A protection order is valuable only if it can be obtained, understood and enforced.
That means monitoring more than how many orders courts issue.
A meaningful system should examine:
- How many applications are made
- How quickly they are processed
- How many are granted or refused
- Whether applicants receive clear information
- Whether police can access the orders
- How breaches are recorded
- How quickly alleged breaches are investigated
- What happens when a person repeatedly violates an order
This is particularly important because a protection order does not physically prevent violence by itself.
It creates a legal mechanism for protection and enforcement.
SAPS has acknowledged that enforcement of protection orders remains inconsistent, with police reporting resource and implementation challenges.
Success would therefore mean measuring whether an order changes the response to danger—not simply whether a document was issued.
Previous violence should become visible before a killing
One of the most important questions in femicide prevention is whether the system recognises repeated or escalating violence.
A killing can be the final event in a longer history of abuse.
That makes previous reports, protection orders, threats, stalking, assaults and breaches potentially important warning information.
A stronger system would allow authorised agencies to identify repeated contact with police, courts, healthcare providers and support services while protecting survivors’ privacy and rights.
The question would be:
When institutions already know that a person is at risk, what happens next?
If multiple warnings exist but remain disconnected across institutions, the problem is not simply a lack of information. It is a failure to turn information into coordinated protection.
Police performance should be measured beyond arrests
Arrests are highly visible and easy to communicate.
They are not, however, the same thing as successful investigation.
South Africa’s SAPS reported that its Family Violence, Child Protection and Sexual Offences units arrested 22,219 suspects and secured 2,941 convictions between April 2024 and March 2025 in cases involving GBV, femicide, sexual offences and crimes against children.
Those figures demonstrate activity, but they do not by themselves establish how effectively the system prevents future violence.
A stronger performance framework would also examine:
- Time taken to investigate cases
- Quality and completeness of case dockets
- Evidence collection
- Forensic turnaround times
- Cases withdrawn and why
- Cases struck off the roll and why
- Cases re-enrolled
- Repeat offending
- Victim and witness protection
- Compliance with domestic-violence procedures
SAPS’s 2024/25 annual report illustrates why this matters. It reported 135,637 contact-crime cases against women pending in court, up from 123,518 the previous year, while also reporting changes in detection and conviction rates.
A growing caseload can reflect more reporting, investigative delays, staffing constraints or several factors simultaneously.
The number needs context.
Prosecutions should be followed all the way through the justice system
There is a similar problem with prosecution statistics.
A case being referred to prosecutors is not the same as a case reaching trial. A trial is not the same as a conviction.
A serious justice scorecard should therefore follow cases through identifiable stages:
report → investigation → arrest where appropriate → prosecution decision → trial → verdict → sentence.
That would reveal where cases are being lost.
If cases disappear primarily during investigation, the response is different from a situation in which investigations are completed but prosecutors lack sufficient evidence.
If cases reach court but repeatedly collapse because witnesses cannot safely participate, survivor and witness support may require attention.
This kind of measurement is more informative than a single national conviction percentage.
Survivor access is itself a measure of system performance
Justice cannot be separated from the services available to survivors.
South Africa’s Thuthuzela Care Centres provide integrated services for victims of sexual offences, including medical care, counselling, police assistance, court preparation and prosecutors. The government describes them as 24-hour facilities.
Success should therefore be measured through access as well as existence.
Important questions include:
- How far does a survivor have to travel?
- Is the service available when needed?
- How long does a person wait?
- Can survivors obtain medical and psychosocial support?
- Is emergency accommodation available?
- Are services accessible to women with disabilities?
- Can migrant women and other vulnerable groups obtain assistance?
- Are survivors treated consistently and with dignity?
- Are services available in rural as well as urban areas?
A centre can be counted as “available” while remaining practically inaccessible to someone who lacks transport, money, childcare or information.
That distinction matters.
The quality of response should be measured by survivors
Government statistics cannot capture everything that matters.
South Africa’s National Strategic Plan explicitly identifies citizen-based monitoring as a way of validating whether progress reported by institutions reflects the experience of people affected by GBVF.
This creates an important additional indicator:
Did the person seeking help experience the system as accessible, safe, respectful and useful?
That can be measured through confidential survivor feedback, independent monitoring and complaints data.
It should not replace administrative statistics.
It should complement them.
A police station might report that every required form was completed while survivors report that they were discouraged from opening cases. Both pieces of information could be important.
Prevention should be measured before the emergency call
If the only indicators begin after violence occurs, the system is measuring response rather than prevention.
South Africa’s national strategy includes prevention and rebuilding social cohesion alongside justice, safety and protection.
That means prevention programmes should also have measurable outcomes.
Depending on the intervention, these might include:
- Changes in attitudes toward violence against women
- Understanding of consent
- Reduction in controlling behaviour
- Safer relationship practices
- Reduced repeat victimisation
- Reduced perpetration
- Greater willingness to intervene safely
- Improved knowledge of support services
Attendance at an awareness event is an activity indicator.
A measurable change in knowledge, attitudes or behaviour is an outcome indicator.
Those are not interchangeable.
A national target needs a baseline
Saying that femicide must “end” provides a direction but not a measurement system.
To evaluate progress, South Africa needs reliable baselines and consistent definitions.
The National Strategic Plan calls for clear, time-bound quantitative and measurable annual and five-year targets.
The same principle applies to individual institutions.
Instead of saying that police response will improve, a monitoring system could publish defined measures for response times, compliance with domestic-violence procedures, unresolved cases and protection-order enforcement.
Instead of saying survivor services will expand, it could measure geographic coverage, waiting times, referrals completed and survivor-reported access.
Instead of saying prosecutions will increase, it could show case progression at each stage.
This makes performance testable.
Data quality is part of the solution
There is a temptation to treat data problems as an administrative issue separate from femicide prevention.
They are not.
If police cannot reliably identify whether victims are women, whether cases involve domestic violence or whether multiple charges relate to the same victim, policymakers cannot confidently identify trends.
SAPS’s decision to withdraw affected disaggregated crime statistics in 2025 demonstrated the consequences of poor data integrity. The service said discrepancies and duplication had affected figures relating to women and children.
Better data requires consistent definitions, accurate recording, interoperable systems and independent quality checks.
It also requires transparency about uncertainty.
A corrected number is more useful than a precise-looking number that cannot be trusted.
The system should measure whether institutions learn from deaths
South Africa has now moved toward a specific mechanism for examining femicide cases.
In August 2026, government launched a Femicide Review Pilot, bringing together government, civil society, researchers and other partners to examine how systems can be strengthened to prevent GBVF and learn from deaths.
This creates an opportunity to ask a different question after every femicide:
Was there a point at which the system could reasonably have acted differently?
That does not mean assuming every killing was preventable.
Some deaths may occur without any prior warning or contact with authorities.
But where there were previous reports, protection orders, threats, breaches or other known risk indicators, a structured review can identify recurring institutional failures.
The purpose should be learning rather than simply assigning blame.
What success would look like in practice
A genuinely successful response would produce several changes simultaneously.
Women at risk would be able to report violence without unnecessary barriers. High-risk cases would receive appropriate and timely responses. Protection orders would be enforceable rather than merely available on paper. Police investigations would progress efficiently. Evidence would move through forensic systems without avoidable delays. Prosecutors and courts would process cases effectively. Survivors would have access to medical care, counselling, shelter and legal assistance.
And over time, the most important outcome should become visible in the population data: fewer women being killed and fewer people experiencing serious and repeated violence.
That last measure should not be expected to move immediately whenever an individual programme changes. Femicide is affected by multiple social, economic, institutional and behavioural factors, so attribution requires careful evaluation.
What should happen when the numbers improve?
Even falling crime figures require interpretation.
A decline in reported GBV could mean less violence.
It could also mean reduced reporting.
A rise in reports could mean more violence.
It could instead reflect greater confidence in institutions.
That is why successful monitoring requires several indicators to move together.
If reporting increases while survivor access improves, investigations become more effective and population surveys eventually show lower prevalence, the evidence for genuine progress becomes stronger.
If reported cases fall while access to police deteriorates, the result is much harder to interpret.
The same principle applies to convictions.
A higher conviction rate can reflect better investigations and prosecutions, but it can also change when the composition of cases entering the courts changes.
Numbers need their denominator, time period and definition.
The measure that matters most is whether the system learns
South Africa already has a framework that calls for prevention, justice, protection, survivor support, coordination and measurable targets.
The challenge is turning those commitments into a transparent chain of evidence.
A speech can announce a target.
A budget can fund a programme.
A police report can record an arrest.
A court can record a conviction.
None of those alone demonstrates that women are safer.
**The meaningful test is whether institutions can show, with reliable data, that risks are identified earlier, responses are faster and more consistent, protection orders are enforced, cases progress through the justice system, survivors can obtain the services they need, and the number of women subjected to lethal and serious violence declines over time. Success against femicide will therefore be measured less by what authorities promise than by what the system can repeatedly demonstrate—and by whether it learns from every case in which a woman is lost.**