Straight answer: yes, cameras you already have can count footfall — no new hardware needed if the entrance view is clear. The number is reliable enough to track trends and conversion, but not precise enough to be treated as an exact headcount.
Every shop owner and gate manager already has a rough sense of how busy yesterday was. What they don't have is a number they can put in a spreadsheet and trust next to last month's. That's the actual gap people counting fills — not "how many people," but "is footfall going up or down, and what's it worth."
Can your existing cameras do this?
Usually, yes. People-counting software runs as analytics on top of a normal CCTV feed — it doesn't need a special sensor. What it needs is a camera angled so that people crossing the doorway don't overlap each other in the frame. A camera mounted straight down over the door, or angled across it from the side, works well. A camera bolted high on a far wall, looking down a long aisle toward the entrance, usually doesn't — by the time people are countable, they're already bunched together.
So the real question at most sites isn't "do we need new cameras," it's "is the entrance camera positioned to count, or just to record." Those are different jobs, and a camera installed for the second one often needs a small repositioning, not a replacement, to do the first.
How accurate is the number, really?
Good enough for a trend line, not good enough for an audit. The software draws a virtual line across the doorway and counts crossings in each direction. Single people walking through, one at a time, at normal pace, get counted reliably. Where it drifts:
- Groups walking side by side can be undercounted if they're close enough to overlap in the frame.
- Trolleys, prams and large bags occasionally get counted as an extra person, or cause a real person to be missed.
- Backlit or glass entrances confuse the software the same way they'd confuse a human glancing at grainy footage — silhouettes are harder to separate than well-lit faces and shoulders.
None of these errors are large individually, and over a week they tend to average out. But if someone asks "exactly how many people came in on Tuesday," the honest answer is "approximately X," not a number you'd defend to the digit.
What is the number actually good for?
Trend, not total. Is this month busier than last month? Is Saturday footfall recovering after the monsoon dip? These comparisons are robust even if the absolute daily number wobbles a little, because the same small errors happen every day.
Conversion rate. Footfall against actual bills or transactions is one of the few retail metrics that tells you something a sales report alone can't — whether a quiet day was low traffic or low conversion. That distinction changes what you'd actually fix.
Staffing and hours. Knowing which hours genuinely draw traffic, versus which hours you're staffed out of habit, is a real operational decision the count supports well.
What it's not good for is a legal or contractual headcount — a mall's common-area footfall dispute, for instance, needs a level of certified accuracy this kind of analytics doesn't promise.
Where does it fall apart?
Two places, honestly. First, multiple entrances that aren't all covered — a count from one door while people also use a side gate or loading entrance is a partial number dressed up as a complete one. Second, a system nobody looks at. A count feeding a dashboard nobody opens is a wasted analytics licence, whichever vendor is running it.
We build this kind of video analytics on existing cameras, so weigh that recommendation with our commercial interest in mind. Our honest advice either way: get the entrance camera positioned correctly before paying for the analytics on top of it — the software can't fix a doorway it can't see cleanly.