Beyond validation: What can connected vehicle data tell us?

For decades, transport agencies have relied on traffic counters, video surveys and roadside monitoring equipment to understand how road networks operate. These methods remain valuable because practitioners understand what the data captures and where its limitations lie.

As CVD becomes more widely available, agencies are increasingly asking whether it can be trusted to support planning, operational and investment decisions. But there is a more useful question: What can connected vehicle data be trusted to tell us?

The value of CVD depends on the decision being made. Transport practitioners should consider what it reveals, what it does not and how it complements other sources of evidence. The focus then shifts from trust alone to selecting the right data for each decision. 

No transport dataset captures every vehicle across an entire network and at all times. An automatic traffic counter (ATC) records vehicle movements at a specific location during a defined survey period. Practitioners can extend the survey period or add locations, but the data remains bounded by where and when it is collected.

CVD offers a different perspective. Rather than observing traffic from the roadside, it captures observations from connected vehicles moving through the network. These generally include newer vehicles and some commercial fleets. CVD does not observe every vehicle, but it can provide information across more locations and over longer periods.

Some observations overlap, allowing agencies to compare and validate the datasets. Others complement one another by revealing different aspects of how drivers interact with the road network.

For example, a roadside survey may provide a detailed picture of conditions at one location over several days. CVD may contain fewer observations at that location but span a longer period and a broader part of the network. Each therefore offers different strengths, and many transport decisions can benefit from both. The key is knowing which source, or combination of sources, can provide the evidence needed for a particular decision. 

To test whether CVD could provide reliable speed insights, we compared it with ATC data collected on Camms Road in Cranbourne, Victoria.

The analysis compared 78,442 vehicles captured by an ATC over nine days with 1,843 connected vehicles observed throughout a full year. Despite the difference in sample size and collection method, both datasets produced similar speed distributions, suggesting they told a comparable story about driver behaviour.

Further analysis examined how sample size affected the stability of speed measurements. Results remained relatively stable as sample size reduced. Below approximately 300 observations, the uncertainty began to affect the usefulness of the results for decision-making. 

Comparing CVD with established data sources helps build confidence, but its purpose is not simply to replicate what an ATC can already tell us.

Because CVD captures observations over time, it can help agencies examine conditions before and after a transport intervention. This can indicate whether behaviour changed, whether the intervention achieved its intended effect and whether that change continued.

Its spatial coverage creates another opportunity. Instead of examining what happened at one survey point, practitioners can explore how speeds change along a road section, how drivers respond to a sequence of road features and where behaviour begins to change or recover. This broader view can show how drivers respond to the combination of several treatments along a single road section. 

Purnell Road, Victoria

Three raised safety platforms were installed along Purnell Road, on either side of an existing roundabout. 

Using CVD, the analysis showed:

• - Vehicle speeds reduced below 60 km/h throughout the road length; 
• - The influence of each treatment extended beyond the installation location; and
• - The interaction between treatments could be observed. 

Unlike a single traffic counter, CVD provided a view of how driver behaviour changed along the road as a whole.

Transport for NSW rail level crossing risk assessment

Transport for NSW manages more than 1,300 public rail level crossings across New South Wales.

The project combined engineering, rail and environmental data with historical crash records and CVD.

The CVD added a new perspective on how vehicle drivers approached and navigated each level crossing, including their behaviour during the journey before reaching it. This enabled a new category of behavioural risk factors to be considered alongside established sources of evidence when prioritising crossings for further investigation. 

CVD gives agencies a way to examine how drivers encounter a road as they travel through the network. It can support questions about how behaviour changes around infrastructure, where a treatment begins to influence speed and how several interventions work together.

Combined with advanced analytics, these insights can help agencies identify patterns, assess risk and make more informed decisions across increasingly complex transport networks.

CVD does not replace established data sources or engineering judgement. It should form part of a broader evidence base, selected and interpreted according to the decision at hand.

When agencies understand what each dataset observes, validation provides a starting point. Combining established evidence with CVD can broaden the questions agencies ask about how drivers interact with the road network and respond to transport investment.

The question is no longer simply whether agencies can trust CVD, but what it can reveal and where it can add value to transport decision-making.