A friend of mine delivered for Uber Eats and DoorDash on an e-bike. Then it was stolen. For a delivery rider, the bike is the job, so the only question that mattered afterwards was simple: would anyone ever be caught?

That question is what this story is about. Crime statistics usually count how much crime happens. This report looks at what happens after a crime is reported: how often Victoria Police identifies someone and takes action, and how often the file is simply left open.

42%
of offences recorded in the year to June 2025 were still unsolved a year later (30% in 2021)
+75%
rise in unsolved offences between the years to June 2022 and 2025
93%
of thefts from cars unsolved, against 2% of drug possession offences
<1%
of cases still unsolved after a year are resolved in the year after that

1. The climb

For most of the last decade, about half of all recorded offences ended in an arrest or a summons to court, and around a third stayed unsolved. Since 2022, that gap has been closing fast.

Source: Crime Statistics Agency (2026a), Table 04, for all lines; open circles from the editions published in 2024, 2025 and 2026 (Crime Statistics Agency, 2024, 2025, 2026a). Years run 1 July to 30 June.

Hover to compare the outcomes in any year. Click Other outcome or Stalled in the legend to add those lines. The open circles show how many offences were listed as unsolved when each year’s data was first published; the arrows show how many of those were solved over the following year.
Text description of this chart A line chart of recorded offences by outcome from the year ending June 2017 to June 2026. Offences ending in arrest or summons stay between about 236,000 and 282,000 a year. Unsolved offences fall to 158,573 during the 2021 lockdowns and 155,113 in 2022, then rise every year to 270,682 in 2025. Open circles show that when each year was first published, the unsolved count was higher: 235,821 for 2024 and 296,080 for 2025. Arrows show these falling by about 21,000 and 25,000 within a year. The 2026 values are a first look, shown with dotted lines.

In the year to June 2025, 270,682 offences (42%) were still unsolved a year after the period ended, up from 30% in 2021. That is a rise of 75% in the number of unsolved offences in just three years.

The open circles reveal something most crime headlines miss: the first year is everything. When the June 2025 figures were first published, 296,080 offences were listed as unsolved. A year later, about 25,000 of them had been resolved, most with an arrest or summons. After that the file barely moves: of the 214,447 offences from 2024 still unsolved after one year, only 1,783 (0.8%) were resolved in the year after that. The newest figures, for 2026, are a first look, so they will improve. But at 45% unsolved, they are only slightly better than last year’s first look of 46%.

2. Two kinds of crime

So which crimes go unsolved? Each bubble below is one type of offence. Press Play and watch how they move over the years.

Source: Crime Statistics Agency (2026a), Table 04. Shows the 48 offence types with at least 500 recorded offences in every year from 2017 to 2025.

Press Play or drag the slider to move through the years. Hover over a bubble for its exact numbers. Click a division in the legend to hide it, or double-click to show only that division.
Text description of this chart An animated bubble chart of 48 offence types from 2017 to 2025. The horizontal axis shows how many offences were recorded and the vertical axis shows the percentage recorded as unsolved. Drug and weapons offences and breaches of court orders (such as a family violence order) sit near the bottom, mostly below 10% unsolved, because police usually identify the offender when they detect the offence. Theft, burglary and property damage sit near the top. Theft from a motor vehicle is the largest bubble, at 93% unsolved in 2025. Over time, most property offences drift upwards and grow larger.

The pattern splits crime in two. When police find a crime, such as someone carrying drugs or breaching a court order, the offender is right there, so the case is solved almost by definition. Only 2% of drug possession offences are unsolved. When a victim reports a crime after the fact, the offender is usually long gone. That is why 93% of thefts from cars stay unsolved.

The crimes that grew fastest after the pandemic, such as thefts from cars and shops and stolen cars, are exactly the ones police are least able to solve.

3. A map of open files

If it all comes down to the type of crime, where you live should still matter, because different places have different crime. Here is the share of each council’s recorded offences that were unsolved in the year to June 2025.

Source: Crime Statistics Agency (2026b), Tables 01 and 05; council boundaries from Australian Bureau of Statistics (2024).

Use the buttons above the map to switch between the unsolved share, the change since 2019, and the overall crime rate. Click Zoom to Melbourne below the map to see the smaller inner-city councils, or scroll to zoom and drag to pan. Hover over a council for its details.
Text description of this chart A map of Victoria’s 79 local government areas shaded by the share of recorded offences that were unsolved in the year to June 2025. The darkest areas are in metropolitan Melbourne, especially the inner, eastern and western suburbs. Maribyrnong has the highest share (55%) and Horsham the lowest (19%). Most regional councils are much lighter. The “Change since 2019” view shows that the unsolved share rose in 61 of the 79 councils. The “Crime rate” view shows that the councils with the most unsolved crime are often not those with the highest crime rates.

The map turns the usual picture of crime upside down. The councils with the darkest shading are not the ones that usually top crime headlines. The three highest are Maribyrnong, Whitehorse and Boroondara, which include some of Melbourne’s wealthiest, lowest-crime suburbs. At the other end, Horsham in the state’s west has only 19% unsolved. Switch to Crime rate and the two maps barely match: across councils, the link between the crime rate and the unsolved share is weak (r = -0.16). Switch to Change since 2019 and most of the state turns red: the unsolved share rose in 61 of 79 councils.

4. It’s the crime, not the postcode

Why would leafy suburbs have worse odds? The answer goes back to Chapter 2. In some councils, most reported crime is theft, burglary and property damage, the kind that is rarely solved. In others, a bigger share is crime that police discover themselves.

Source: Crime Statistics Agency (2026b), Tables 01, 02 and 05. Population is estimated from the CSA's offence count and rate per 100,000.

Find your council: type its name in the search box above the chart, or click any bubble to highlight it (double-click to clear). Use the tick boxes to compare police regions.
Text description of this chart A scatter plot of 79 councils. The horizontal axis is the share of each council’s offences that are theft, burglary or property damage, from about 30% to 70%. The vertical axis is the share unsolved. The points form a tight upward line (correlation 0.96): councils where property crime makes up more of the total have more unsolved crime, whatever their region or size.

The points fall on an almost straight line. Across 79 councils, the share of crime that is theft, burglary or damage accounts for about 92% of the variation in unsolved rates (r = 0.96). This does not prove one causes the other, but it is a far simpler explanation than “some police stations try harder than others”. In councils like Maribyrnong, more of the crime reported is the kind that leaves no suspect behind.

5. What are your odds?

So, back to my friend’s e-bike. Bike theft sits near the very top of the list: in the year to June 2025, 87% of bike thefts were still unsolved a year later. And because almost nothing is solved after the first year, that figure is close to final. Here are the recorded outcomes for the crimes most likely to happen to an ordinary Victorian.

Source: Crime Statistics Agency (2026a), Table 04. "Other outcome" includes cautions, pending summonses, withdrawn complaints and other statuses.

Switch between 2025 and 2019 with the buttons to see how the odds have changed. Hover over a bar segment for the exact share.
Text description of this chart A stacked bar chart showing the outcomes of 11 everyday offence types in the year to June 2025, sorted by share unsolved. Something stolen from your car has the largest unsolved share (93%). Assaults and robberies have the lowest unsolved shares. Switching to 2019 shows smaller unsolved shares for 9 of the 11 offence types.

So, should you still report it?

Four things to take away

  1. More crime is going unsolved. 42% of offences recorded in the year to June 2025 were still unsolved a year later, up from 30% in 2021.
  2. The type of crime decides the odds. 93% of thefts from cars stay unsolved, against 2% of drug possession offences.
  3. It’s the crime mix, not the postcode. Leafy suburbs have some of the worst odds, and the likely reason is that more of their crime is theft and damage (r = 0.96 across 79 councils).
  4. The first year is everything. If a case is not resolved within about a year, it almost never is: fewer than 1 in 100 are resolved in the year after that.

Yes, you should still report it. A report is the only way a crime gets into these statistics at all. Without the people who filed reports they knew would probably go nowhere, this story could not be told.

What the numbers change is the question worth asking. Crime will be an issue again when Victorians vote on 28 November. Most of the debate will be about how much crime there is. This data suggests a sharper question: how much of it gets solved? Right now, for the crimes most of us will actually experience, the honest answer is: not much, and less than a few years ago.

Case closed? For most stolen bikes and car break-ins in Victoria, the case never really opens.

About the data

What “unsolved” means. The Crime Statistics Agency (CSA) records each offence’s investigation status as Victoria Police held it when the data was extracted, in July after each year ends. “Unsolved” means no outcome has been recorded. “Stalled: not enough information” is the CSA status Not authorised: a crime occurred, but there was not enough information to proceed at the time. “Other outcome” groups cautions and official warnings, pending summonses (Intent to summons), withdrawn complaints and other statuses (Crime Statistics Agency, n.d.).

Why most charts use 2025, not 2026. Investigation status is a snapshot that changes as cases are solved. Comparing three editions of the same CSA table shows that unsolved counts fall noticeably in the first year after publication (by about 21,000 offences for 2024 and 25,000 for 2025), then barely change (0.8% in the second year). The latest year, 2026, has not had this first year of investigation, so it would make the picture look worse than it will be. Charts 2 to 5 therefore use the year to June 2025, the most recent year with at least 12 months of follow-up, and Chart 1 shows 2026 separately as a first look.

Limitations.

  • Recorded crime only includes offences reported to or found by police. Many crimes are never reported.
  • An arrest or summons is not a conviction.
  • In the 2026 first look, 28,044 offences were awaiting a summons, far more than in earlier years, which may also reflect a change in how statuses are recorded.
  • The council-level data groups outcomes into four statuses, so Charts 3 and 4 use fewer categories than Charts 1, 2 and 5.
  • Chart 4 shows a correlation between councils, not a cause, and it does not describe individual cases.
  • The ABS boundaries approximate council areas using mesh blocks. They are used for display only.

Design choices. The report uses a martini-glass narrative structure (Segel & Heer, 2010): each chapter opens with a headline that states the finding, annotations point to the key moments, and interaction then lets readers explore the details on demand, such as finding their own council. Red always means unsolved and blue always means arrest or summons, taken from ColorBrewer’s colour-blind-safe RdBu palette (Harrower & Brewer, 2003), so the encoding carries across all five charts. The map uses ColorBrewer sequential and diverging scales, and categories use the colour-blind-safe Okabe-Ito palette (Wong, 2011). Colour is never the only cue: hover labels name every element, key points are annotated, and every chart has a text description. The interactive charts were built in R with plotly and crosstalk (Sievert, 2020).

Processing. All data was downloaded directly from the CSA and the ABS and processed in R. CSA statuses were grouped into four outcomes, and shares were calculated from offence counts. Council property-crime shares combine the CSA subdivisions B20 Property damage, B30 Burglary/Break and enter and B40 Theft. Council boundaries were re-oriented so that plotly draws them correctly. The full code is included with the submission.

References

Anthropic. (2026). Claude (Opus 5.5) [Large language model]. https://claude.ai

Australian Bureau of Statistics. (2024). Local Government Areas: Australian Statistical Geography Standard (ASGS) Edition 3, 2024 [Data set]. https://geo.abs.gov.au/arcgis/rest/services/ASGS2024/LGA/MapServer

Crime Statistics Agency. (n.d.). Glossary and data dictionary. Retrieved October 27, 2026, from https://www.crimestatistics.vic.gov.au/about-the-data/glossary-and-data-dictionary

Crime Statistics Agency. (2024). Recorded offences: Year ending June 2024 [Data set]. Victorian Government. https://files.crimestatistics.vic.gov.au/2024-09/Data_Tables_Recorded_Offences_Visualisation_Year_Ending_June_2024.xlsx

Crime Statistics Agency. (2025). Recorded offences: Year ending June 2025 [Data set]. Victorian Government. https://files.crimestatistics.vic.gov.au/2025-09/Data_Tables_Recorded_Offences_Visualisation_Year_Ending_June_2025.xlsx

Crime Statistics Agency. (2026a). Recorded offences: Year ending June 2026 [Data set]. Victorian Government. https://files.crimestatistics.vic.gov.au/2026-09/Data_Tables_Recorded_Offences_Visualisation_Year_Ending_June_2026.xlsx

Crime Statistics Agency. (2026b). Recorded offences by local government area: Year ending June 2026 [Data set]. Victorian Government. https://files.crimestatistics.vic.gov.au/2026-09/Data_Tables_LGA_Recorded_Offences_Year_Ending_June_2026.xlsx

Harrower, M., & Brewer, C. A. (2003). ColorBrewer.org: An online tool for selecting colour schemes for maps. The Cartographic Journal, 40(1), 27–37. https://doi.org/10.1179/000870403235002042

Segel, E., & Heer, J. (2010). Narrative visualization: Telling stories with data. IEEE Transactions on Visualization and Computer Graphics, 16(6), 1139–1148. https://doi.org/10.1109/TVCG.2010.179

Sievert, C. (2020). Interactive web-based data visualization with R, plotly, and shiny. Chapman and Hall/CRC. https://plotly-r.com

Wong, B. (2011). Points of view: Color blindness. Nature Methods, 8(6), 441. https://doi.org/10.1038/nmeth.1618

Generative AI acknowledgement

I acknowledge the use of Claude (Anthropic, 2026) in this assignment. I gave Claude the assignment brief, the marking rubric and my feedback from earlier assessments, and asked it to suggest unique story ideas using open data. I chose the “Case Closed?” story from its five suggestions and provided the personal example used in the introduction. Claude then analysed the Crime Statistics Agency data, wrote the R code for all five visualisations, drafted the report text, and tested and checked the report against the rubric.

Prompts I used included (lightly edited): “give me 5 top options and then I will choose one”, “I want to go with Case Closed”, “opening line can be like my friend used to do Uber DoorDash delivery from e-bike and it got stolen” and “check again all the rubric markings and feedback and fix any weak issues”.

All figures in the report are calculated by the R code directly from the original data sources listed in the references.