Do We Need New Traffic Data?

You have a traffic question and a folder of old surveys, camera footage or dashboard results. Is another survey necessary?

Use this guide before approving data work. Choose between reusing a result, analysing a source you already hold, and collecting something missing. Leave with a recommendation your manager or consultant can act on.

The four-page PDF contains a reuse check, a data-selection guide, three worked decisions and a decision record you can type into or print.

1. Write the question the data must answer

Name the measure, location and period. Then name the decision it will inform.

“We need traffic data for the town centre” is too open. Try: “Which of three town-centre approaches has the longest weekday afternoon journey time, to prioritise a detailed congestion study?”

That question needs comparable journey times. It does not yet need a fresh count on every approach. If the intended decision is unclear, clarify it with the decision owner before selecting data.

2. Can you reuse what you have?

Start with the source files behind previous reports, council survey registers, road-authority counters, signal records, licensed datasets and retained footage. A report summary may hide useful detail in its original files.

Check these five things against your question. Write yes, no or unknown, with a short reason; do not add up a score.

1. Right measure and place? The source records the variable you need at the relevant section, direction, lane or movement. An approach total does not identify individual turns.
2. Right period and conditions? The days and times match. Check changes to access, traffic control, land use, routes, works and school or event activity since collection.
3. Enough usable coverage? Inspect valid intervals, sample sizes, missing periods and who is represented. A blank interval is not zero traffic. A large vehicle sample does not measure people walking.
4. Traceable and checked? You can identify the owner, method, units, time zone, location map and quality checks. Read the limitations behind a dashboard value.
5. Permitted and available? You can obtain and use the source for this purpose. Check licence, footage retention, access and handling requirements before relying on it.

All five are yes: reuse the evidence, or analyse the underlying files if the answer has not yet been extracted. No new collection is needed for this question.

A check is unknown: ask the owner for the missing information first. Unclear metadata is a clarification task, not an automatic reason to buy another survey.

A check is no: identify whether that gap could change the decision. If it could, fill that specific gap. If it could not, record why the evidence is still sufficient for the stated use. Data that fails for a current decision can remain useful historical context. Unauthorised use must be resolved before proceeding.

How old is too old?

This guide sets no expiry age. An older count on an unchanged road can remain useful for initial screening; last week's count during a diversion may be unsuitable for ordinary conditions. The required precision and any applicable authority requirements also matter. Record what has changed and whether it affects the question.

3. Match the gap to the data

Choose the measure first, then the source. Each option includes a check that can prevent a poor purchase.

Vehicle totals, classes or speeds at a point

Use counts or a point-speed survey. First look for a suitable existing counter or survey. Where collection is needed, choose tubes, radar or another validated method suited to the site and measures.

Check direction, count intervals, vehicle classes and speed definitions. Request the distribution and sample size when speed variation matters; a single average hides it. A point count cannot establish where trips started or ended. Road-authority datasets may already contain volume, class and speed bins; inspect their stated limitations.

Journey time, route choice or corridor patterns

Consider connected-vehicle or probe data: observations from a subset of vehicles. It can help compare routes or time periods without installing equipment at every location, if the provider supplies the required coverage.

Ask for sample counts by link and period, missing or suppressed intervals, vehicle mix and how journeys are matched to roads. Distinguish point speed from journey speed that includes stops. Do not treat the sample as a total traffic count without a validated estimation method.

Turns, crossings, queues or interactions already on camera

Analyse retained video before arranging another recording. Check a sample from the actual periods: can you see every required movement, read the timestamps and follow activity without obstruction?

Define counting lines, classes and any interaction measure before analysis. Request a manual spot-check of the output. Footage does not automatically provide reliable speed measurements; that requires a suitable validated method. A short recording with no observed conflicts does not establish that a site is safe.

Current activity that existing sources missed

Use temporary video or sensing when the required movements, users or periods are absent from the record. This includes temporary XVision sensing where its proposed configuration can measure the gap.

Before deployment, agree the view, outputs, observation period and quality checks. Test visibility at relevant lighting and traffic conditions. If the question needs only vehicle counts or point speeds, consider the simpler survey above.

Visibility, parking behaviour or the reason for a manoeuvre

Start with a site visit or structured manual observation. Record geometry, obstructions, kerb use and the conditions when the behaviour occurs. A photograph can document an obstruction; parking occupancy needs repeated observations over the relevant period.

Separate what you saw from what you inferred. Use repeated observation or video if the decision depends on how often something happens. One visit cannot establish a typical weekly pattern.

Enough data, but no defensible conclusion

Commission interpretation rather than collection. Give a traffic engineer or analyst the source files and decision question. Ask them to reconcile conflicting measures, test assumptions and state what can be concluded.

If further evidence is needed, require them to identify the precise gap before proposing collection. Reprocessing a dataset and purchasing a new survey are separate tasks.

Useful context that is not a substitute

Crash records and community reports help identify locations and patterns to investigate. Neither supplies a missing count, speed distribution or current exposure on its own. Signal records can be useful too: check the detector-to-lane mapping before using them for a movement question. For example, Victoria's public SCATS data contains counts by detector and 15-minute interval, not a ready-made turning survey.

4. See the choice in practice

These examples are fictional. The periods illustrate the question, not a prescribed survey duration.

Reuse: choose an hour to investigate

Question: Which weekday afternoon hour carries the most westbound vehicles at the council depot entrance?

Available: Original 15-minute directional counts from a recent representative survey at that entrance. Valid intervals, method and permissions are confirmed; access and operating conditions have not changed.

Decision: Reanalyse the existing intervals and report variation between surveyed days. Do not order another count to answer the same question. The result selects an observation window; it does not establish annual traffic or explain congestion.

Analyse: a report omitted the movement

Question: How often does a queue block the service-lane exit between 4 and 6 pm?

Available: The old report contains approach totals only. The original footage is still held, covers the exit and required days, and has verified timestamps and authorised use.

Decision: Analyse queue-blocking episodes and duration from that footage, with a manual check. The missing evidence is an output, not a missing recording. If the view is obstructed, collect only the missing view or period.

Supplement: corridor data cannot see an individual turn

Question: Which left-turn movements coincide with people crossing the side-road entry during the morning peak?

Available: Connected-vehicle journey times and a council camera that cannot see the crossing. Neither source identifies the interaction in question.

Decision: Collect targeted, simultaneous observations of the turn and crossing, using a suitable video view or structured manual study. Retain journey times as corridor context. Check that the selected period captures the relevant activity before drawing a conclusion.

5. Leave one clear recommendation

Complete the last page of the PDF: state the question, list the sources checked, record the decisive gap and select reuse, analyse, clarify or collect. Include the result that would be sufficient and who will review it. This makes “no new survey” as reviewable as a request for new work.

For any new collection, take the selected measures and coverage into the Traffic Survey Scope of Works Template. That resource handles the brief; this guide handles the choice of evidence.

If you want XVision to assess an evidence gap, bring the completed record to Investigate a Location. Start with the question and source descriptions; arrange authorised access before sharing footage.

Basis and limits

The decision sequence and examples are XVision's practical synthesis, not a scoring standard or an authority approval. Apply the requirements of the intended decision when selecting a method and collection period.

For camera material, check the owner's authority and applicable privacy rules. OAIC's camera guidance explains federal coverage and points to state and territory laws; it does not establish a council's permission to reuse footage.

References

• Austroads, Guide to Traffic Management Part 3: Transport Studies and Analysis Methods — survey planning, sampling, data integrity and methods.
• Victorian Department of Transport and Planning, TIRTL Traffic Counts and Classification — available measures and dataset limitations.
• Victorian Department of Transport and Planning, Traffic Signal Volume Data — detector counts and aggregation.
• Aljamal, Abdelghaffar and Rakha (2020), Real-time Estimation of Vehicle Counts on Signalized Intersection Approaches Using Probe Vehicle Data — estimation methods and sensitivity to probe coverage.
• OAIC, Security cameras — privacy coverage and links to state and territory requirements.