Australian Foot Traffic Data: Who Actually Has Coverage, and Do You Even Need It? (2026)

Most of what’s written about foot traffic data is written for Americans. That matters, because the providers everyone names (Placer.ai, SafeGraph, Unacast, Foursquare) were built for a country with 340 million people and a very different retail landscape.

Australia has about 27 million people, proportionally fewer phones in these datasets, and a lot more of our shops sitting inside enclosed centres and under office towers. None of that makes the data bad. It’s the same technology working here as anywhere. It does mean the numbers need reading differently, and that the right source depends heavily on the question.

This guide covers who actually has Australian data, what each source is genuinely good at, and, first, whether you need it at all.

Start here: does foot traffic even matter for your project?

Foot traffic only matters if your business converts people who are already walking past. Some do. Many don’t.

Passing trade businesses live and die on footfall. A coffee shop, a takeaway counter in a food court, a phone repair kiosk, a convenience store. If nobody walks past, nobody comes in.

Destination businesses generate their own traffic. People decide to go there and travel to it. A childcare centre, a gym, a furniture showroom, a medical practice, a Bunnings. Existing footfall on the street tells you almost nothing about how these will perform, because the customers weren’t going to be there anyway. What matters is who lives and works in the catchment, and how easily they can drive to you and park.

Most QSR sits in between, and the split depends heavily on format. A CBD lunch site is close to pure passing trade. A drive-thru on an arterial is close to pure destination and vehicle exposure. The same brand can need completely different data for two sites.

If you’re a brand or franchisor

Which side you sit on is measurable, not a matter of opinion. It can be quantified from your own trading history, and we come back to how at the end of this guide. For a lot of brands the answer lands lower than expected. For some it lands higher.

Knowing which applies before you sign matters, because the alternative plays out the same way every time. A site presents strong footfall numbers. The rent is priced to match those numbers. Then the trade never eventuates, because the people walking past were never the target market to begin with. You end up carrying premium rent for an audience you were never going to convert, on a lease with years left to run.

It’s one of the most common and most expensive mistakes destination-based businesses make, and it’s usually visible in your own network’s numbers well before you commit to the site.

For property developers specifically

Three traps worth naming.

High footfall means high rent. If your use doesn’t monetise passing trade, you are paying a premium for something you can’t convert. A busy pedestrian strip is a cost, not a benefit, for a tenant whose customers all arrive by appointment.

Footfall composition beats footfall volume. A Brisbane CBD strip records enormous weekday numbers and empties on weekends. Surfers Paradise or Byron records enormous numbers of people who will never return. Ten thousand people walking past once each is a completely different business to two thousand walking past five times a week, and on most footfall dashboards those two sites look similar.

Ground-floor retail in a residential or mixed-use development is usually a catchment question, not a footfall question. The tenants who will actually sign are serving the building and the surrounding blocks. Counting existing passers-by on a street that’s about to change fundamentally tells you little.

The test that applies either way

If you can explain it, read on, but read the next section first, because the Australian numbers come with conditions attached.

How the numbers are built, and why sample size matters more here

Every one of these providers does the same three things. They collect location signals from phones where the owner has agreed to share location through an app. They match those signals to a business. Then they scale up from that sample to estimate the total.

That last step is where Australia gets harder.

Say a provider has 5% of all phones in its sample. In the US that’s about 17 million devices. In Australia it’s about 1.35 million, for the whole country.

Now spread those out. Greater Sydney might get 275,000. A regional area with 8,000 residents gets around 400. If a café there is visited by 3% of them on a given day, the provider is working from roughly a dozen observations before any quality filtering.

Then they multiply that by twenty to produce the estimate you see.

The result looks identical to a Sydney number: a clean figure with a trend line. It carries far more uncertainty, and nothing on the screen says so.

Who actually has Australian coverage

The international names

Placer.ai: no Australian foot traffic coverage. The best-known platform, and it doesn’t operate here. Their reports circulate widely in Australia and you’ll see them quoted. If someone is showing you Placer numbers for an Australian site, ask exactly what they bought.

SafeGraph / Dewey: places yes, visits no. You can potentially buy Australian business locations from them. The visit counts are a US product.

Unacast: available, density unknown. They sell Australian data through global feeds, but describe their international coverage as varying by country. Worth pursuing only if they’ll tell you how many devices they actually observe in your specific area.

Foursquare: good locations, marketing-focused visits. Genuine Australian coverage of business locations. The visitation products are built to answer “did our campaign work?” rather than “should I sign this lease?”, and the data skews young and inner-city.

StreetLight (Jacobs): real Australian presence, but it counts vehicles. They have an Australia and New Zealand business with local government clients, including work with the City of Gold Coast on the Surfers Paradise Esplanade and network analysis for Brisbane. Excellent for traffic volumes past a site. It does not tell you who walks in the door.

Azira: Sydney office, check the product. Formerly Near. Operating here with broad international location coverage, but their newer packaged visitation product launched covering the US and Canada. Ask which product carries Australian data.

The Australian sources

Beyond the international platforms, there are several categories of Australian data worth knowing about, and for many questions they’ll serve you better than a global panel will.

Telco network data. Several Australian datasets are derived from mobile network activity rather than from apps. That distinction matters: it doesn’t depend on whether someone installed a location-sharing app, which gives these sources much stronger regional representation than any app-based panel operating here. The trade-off is that network positioning is coarser, so they’re excellent for movement patterns and catchments and unsuited to individual shops.

Transaction data. Bank and payments datasets cover a far larger share of the population than any phone-based sample, and they identify the merchant exactly rather than inferring it from geography, which sidesteps the multi-tenancy problem entirely. The limitation is that they only see purchases, so browsing, window-shopping and cash trade are invisible.

Sensor and counter networks. Wi-Fi, camera and radio-frequency counters installed in centres and individual stores. This is genuine measurement rather than estimation, which makes it the closest thing to ground truth available. The catch is access: whoever owns the venue owns the data, so landlords hold it and prospective tenants generally don’t unless it’s shared during negotiation. Always ask what counter data exists for a centre you’re considering.

Modelled traffic datasets. Australian transport modelling firms produce national road and pedestrian traffic layers, licensed directly and also sitting underneath several commercial platforms, so you may already have access to one without realising. The method sets the local benchmark: mobile data is grouped into trips, then checked against physical traffic counts collected by transport agencies at thousands of Australian sites. That checking step against real counts is the standard worth holding any source to.

Worth asking of any platform: where does the data actually originate? Plenty of good products license their underlying sample from someone else, which is perfectly reasonable, but it matters if you’re running two tools side by side. If both draw on the same source, agreement between them isn’t corroboration. Ask before you treat a second subscription as a second opinion.

Three things worth understanding before you lean on the numbers

1. Regional numbers lean more on the model than the sample

Around 28% of Australians live outside the capital cities, spread across a landmass the size of the continental US. The phone-based providers do not have meaningful numbers of devices out there. What they have is a formula, built on how metropolitan areas behave, applied to places that don’t behave that way.

What to do: before accepting any regional estimate, ask how many devices were actually observed in that specific area. Not national coverage. Not “yes, we cover Australia.” The local count. If that isn’t available, treat the number as a model output and weight your decision accordingly.

2. Individual shops inside centres are hard to separate

This is the biggest issue in the Australian market, and it’s harder here than in the US because so much more of our retail sits inside enclosed centres and under towers.

Phone GPS is accurate to roughly 5 to 15 metres outdoors. Inside a building, surrounded by concrete and steel, that gets much worse, often 20 to 50 metres. And critically, GPS can’t tell what floor someone is on. A phone on level 3 of a centre and a phone in the basement car park directly below register in almost the same place.

What that means in practice:

• Two shops with 6-metre frontages sit well within a single margin of error. Visits attributed to your site may belong to the neighbour.
• Ground-floor retail, level-one retail, the offices above and the car park below all share one footprint. The provider is estimating between them.
• Car parks generate enormous dwell time that has to be assigned somewhere.

If a provider gives you a single coordinate rather than an outline of the actual tenancy, shop-level numbers aren’t just imprecise. They aren’t really available.

Australia does have good address infrastructure. G-NAF is a national, freely licensed address file including unit and level numbers, and it’s better than the American equivalent. Registered leases appear on title in our Torrens states. But neither gives you a searchable national list of who trades from which shop, and even if it did, it wouldn’t fix the floor problem.

What to do: use foot traffic at centre level, not shop level. Total centre visitation and entry-point counts from the landlord are defensible. “This tenancy received 4,200 visits last week” is not, and shouldn’t be the deciding input in a lease.

3. “Visitor demographics” are usually inferred, not measured

Nearly every provider sells you a profile of who visits. Here’s how it’s actually made: they work out where a phone spends its nights, assume that’s home, match it to a suburb, then apply that suburb’s census averages to the person.

So it isn’t a measurement of your visitors. It’s an estimate of where they live, followed by an assumption that they’re typical of that area. Within any given suburb, incomes and household types vary enormously, often more than they vary between suburbs.

Transaction-based datasets are the exception. There, the customer attributes are real rather than borrowed from a suburb average, and the merchant is identified exactly rather than inferred from a coordinate.

The other half of the picture: transport models

There’s a whole category of analysis that works differently, and it’s under-used in retail property. It isn’t a replacement for mobility data. As you’ll see, it runs on mobility data. But it puts that data to a different and generally more powerful use.

A transport model builds a simulated population of an area from census and household survey data, gives each simulated person a realistic set of attributes and daily activities, and routes their trips across the actual road and transit network. Australia has several mature examples, maintained by state road authorities and by specialist transport modelling consultancies.

What the modelling adds:

Demographics that are structural rather than inferred. Each simulated person carries age, household type, income and car ownership because they were constructed that way from census and survey data, not because a device was observed in a particular postcode overnight.

It knows why the trip is happening. A device passing at 8:10am on the way to work and one passing at 11:30am on a shopping trip look the same in raw mobility data. They’re completely different customers, and trip purpose is what separates them.

It understands routes, not just radii. People shop on the way home from work. Two centres can be 3km apart and barely compete because they sit on different commute routes, something a radius-based analysis can’t see, and something that matters a great deal when you’re worried about a new store cannibalising an existing one.

Regional coverage holds up better. This is the significant one for Australia. A model’s accuracy in regional Queensland doesn’t rest solely on how many devices are observed there. It draws on the travel survey, the road network and physical traffic counts as well, all of which exist outside the capitals, and all of which help where the device sample is thinnest.

It can answer “what if?” Mobility data describes what already happened, which it does very well. A model can also tell you what changes when the bypass opens, the competitor builds, or the 400-lot subdivision settles. Property decisions are forward-looking, so you generally need both.

Where mobility data becomes most valuable

This is the part usually missed in the “which provider should I buy?” conversation. Mobility data isn’t in competition with transport modelling. It’s one of the three inputs that make a modern model work, and it’s the input that has improved most in the last decade.

A well-built model calibrates against three sources at once, and each one covers the others’ weaknesses:

• Household travel surveys such as VISTA, QHTS and PATHS give you rich detail on who travels, why, and in what sequence. Small samples, usually a single travel day, and expensive to run.
• Mobility data gives you scale and continuity no survey could ever afford: millions of observed movements, across every day of the year, at times of day and in corridors a survey would never capture.
• Physical traffic and pedestrian counts anchor the whole thing to reality at specific points on the network.

Used together, they produce origin and destination insight that none of them delivers alone. The survey supplies the who and the why. The mobility data supplies observed volume and timing at a scale that lets you see genuine travel patterns rather than surveyed intentions. The counts hold the result honest. Mobility data used this way, as a calibration and validation layer inside a structured model, is doing considerably more work than the same data used on its own to produce a visit count for a single address.

Where the modelling is weaker: travel surveys sample a small share of households on a single day and under-record short local trips, which is exactly the sort retail depends on. Transport planners specify shopping trips fairly crudely, since their concern is traffic loading rather than which shop. And a model doesn’t observe anything directly, so a wrong assumption produces a confident wrong answer with nothing to flag it.

The practical takeaway: for a single site, in a strong metro location, with a passing-trade concept, a good mobility dataset may be all you need. For regional sites, network planning, cannibalisation questions, or anything forward-looking, the same data is far more valuable feeding a calibrated model than standing on its own.

Which source for which question

How to read a foot traffic number someone hands you

Most people never buy this data directly. It arrives inside something else: a leasing pack from the landlord, a broker’s submission, a consultant’s report, a franchisee’s own research. By the time you see it, it’s a figure in a table with no working shown.

Six questions to put to whoever supplied it:

1. Where did this actually come from? Name the underlying source, not the platform it was displayed in.
2. Is this the centre, or is it this tenancy? If it's the tenancy, how was it separated from the shops either side?
3. How many devices were genuinely observed here before anything was scaled up?
4. What period does it cover, and is that a normal trading pattern for this location?
5. Is it a count or an estimate? Sensor and counter data are counts. Everything else is modelled.
6. What's the range around it?

None of this is about catching anyone out. A modelled estimate is perfectly legitimate, and often it’s the only thing available. The point is that a figure with a known source and an honest range is evidence you can weigh, and a figure without either is decoration.

The measurement almost nobody runs: what is foot traffic actually worth to you?

Everything above answers one question. How many people are there.

Not one of these sources answers the question a lease decision actually turns on, which is how much those people are worth to your particular business. Those are different questions, and the second one is answerable. It just gets answered from your data rather than a provider’s.

If you already trade from more than a handful of sites, the answer is sitting in your own numbers. Sales by site, matched against the characteristics of each location: passing traffic, catchment size and composition, drive times, competition, visibility, parking, co-tenancy, format. Model that properly and the variation can be attributed. You find out how much of the gap between your strongest and weakest sites is explained by footfall, and how much by everything else.

The result is specific to your brand, and often to your format. Two operators in the same category can land in very different places. The same brand can land in different places for a drive-thru and a CBD store.

Sometimes it comes back higher than expected. Footfall really is the driver, the busy site really is worth the rent, and better data on it is worth paying for. That’s a useful thing to have confirmed rather than assumed.

More often, at least for destination formats, it comes back lower. Passing traffic explains less of your performance than catchment composition or drive-time access does. If that’s your situation, you have been paying a rent premium for a variable that barely moves your revenue, and you now know to stop.

Either answer changes what you do next, because it tells you which variables to spend money measuring accurately and which you can afford to hold loosely. If footfall explains a small share of your sales variation, a wide band around a regional estimate is perfectly tolerable, and most of the caution in this guide matters less to you. If it explains a large share, precision is worth paying for, and those six questions become the ones you ask on every deal.

This is where SiteGaps IQ sits, and it’s deliberately not another foot traffic feed. We take the data described in this guide, clean it, test it, and measure what it is actually contributing to your revenue, so the weight you put on it comes from your own trading history rather than from a broker’s assumption or a dashboard’s confidence.

The bottom line

Come back to the question this guide opened with, because it decides everything downstream. If you can’t trace the path from someone walking past to someone spending money with you, no provider here is selling you anything you need.

If you can trace that path, foot traffic data is worth having, and it’s worth understanding what you’re holding. In most Australian cases it’s a well-constructed estimate rather than a count, and inside a centre or a multi-level building it can’t reliably separate your shop from the one next door. Neither of those is a reason to dismiss it. They’re reasons to ask for a range, and to pair it with the catchment and access data that answer the questions it can’t.

A number with an honest range attached is a decision you can defend at a board meeting or a franchise council. “Between 2,800 and 6,500 a week, and we can’t isolate this tenancy from its neighbours” is less comfortable to present than a flat 4,200, and considerably more useful, because it tells everyone in the room how much weight the figure will bear.