Behind the Counter

Shoplifting at Victorian shops is up 77% since 2017. Crimes against people there rose 54%.

Hover over any year to compare all three lines. Switch to Number of offences to see the scale behind the growth.

After the lockdown dip, both lines climbed past their pre-pandemic levels. In 2025, police recorded more than 8,000 crimes against people at Victorian retail locations, roughly 22 every day. Meanwhile deception fell by half as cash gave way to cards.

Crimes against people rose fastest at 24-hour convenience stores: up 103% since 2017.

Each row is a type of shop; each square is a year. Orange means more offences than in 2017, purple means fewer. Bars on the right show this year’s volume. Switch to Shoplifting to compare.

Shopping centres see the most incidents, but the fastest growth was at 24-hour convenience stores, where crimes against people rose 103%. Supermarkets rose +71%, while service stations barely changed (-6%). Milk bars recorded fewer incidents, probably because fewer milk bars are left.

Shop crime peaks in two kinds of place: inner Melbourne and regional centres like Latrobe and Greater Shepparton.

Darker councils have more offences at retail locations per resident. Hover for details, use Zoom to Melbourne to see the metropolitan councils, and search or click a council to highlight it here and in the next chart. Double-click to clear.

4 of the 10 councils with the highest rates are outside Melbourne. These are regional centres, where one shopping strip serves a whole district.

It isn’t simply a poverty story. Hotspots sit at both ends of the disadvantage scale.

Each circle is a council: further right is less disadvantaged, higher up means more offences at shops per resident, and bigger circles mean more offences in total. Click a circle to find it on the map above. Click a legend item to hide that group.

Excluding the CBD, the relationship between disadvantage and the rate is weak (r = -0.21): more disadvantaged councils tend to have higher rates, but with plenty of exceptions. The high-rate councils fall into two groups: affluent inner-city shopping strips that draw crowds from across Melbourne, and disadvantaged regional centres where much of a district’s shopping happens in one place.

Police kept pace: every year, 45–51% of shoplifting reports end in an arrest or summons.

Each bar is a year of shoplifting reports, split by what happened next. Numbers above the bars are total reports. Switch to Number of offences to see the same shares applied to a much bigger total. The hatched bar (*) is the most recent year, which is provisional.

The share of reports ending in an arrest barely moved while the number of reports nearly doubled. The surge is not a story of police giving up. It is a story of volume, and that volume is met first by whoever is standing behind the counter.

So what?

The shoplifting surge is real, but it is only half of what the data shows. Since 2017, crimes against people at Victorian shops have risen 54%, fastest at 24-hour convenience stores, and they cluster in inner Melbourne and in regional centres. Police have kept pace on shoplifting, solving about the same share each year. What has not changed is who meets it first. The next time a headline counts stolen goods, like the two slabs that walked out of my shop that Friday, it is worth asking who was standing at the register.

A note on the data

The Crime Statistics Agency records where an offence happened, not who the victim was. “Crimes against people at retail locations” includes offences against customers as well as staff. Figures are offences recorded by Victoria Police, so changes in reporting can affect trends: Victoria Police has noted that new online reporting has made retail theft more visible (Australian Associated Press, 2026). Council rates are per resident, which overstates rates in places that draw many visitors, especially the Melbourne CBD. Disadvantage scores are from the 2021 Census. Metropolitan councils are the 31 councils of metropolitan Melbourne. Years run July to June. The most recent year is provisional and may be revised.

Acknowledgements

I Acknowledge Generative AI (ChatGPT) was used to assist with understanding technical aspects of R and RPubs, reviewing spelling and grammar, and identifying relevant sources of information. All suggestions and sources were independently reviewed and verified. The research, analysis, code, and final assignment content were completed by me.