People counting is a solved problem. Occupancy analytics is a solved problem. Neither tells you whether anything happened. This is the distinction between measuring presence and measuring behavior, and why retail, venues and events keep buying the first while needing the second.
Counting is genuinely good at a specific set of questions: how many people are in this space, is it approaching capacity, when are the peaks, how does this Tuesday compare to last Tuesday. For safety, staffing and opening hours, that is the correct tool and there is no reason to over-engineer it.
The trouble starts when counting gets asked to justify commercial decisions it was never built to support.
Imagine three visitors, each stationary in the same square metre for ninety seconds. One is studying a product. One is waiting for someone. One is lost and reading a sign.
Every counting and occupancy system on the market records these three identically. If you redesign your space on that basis, you will optimise for confusion and queuing, because those produce the highest dwell numbers in the building.
Dwell time without behavioral context is not a quality metric. It is a mixture of interest, friction and navigational failure, and the three move in opposite commercial directions.
| Question | Presence data | Behavior data |
|---|---|---|
| How busy was it? | Answers it | Answers it |
| When are the peaks? | Answers it | Answers it |
| Which zone performs best? | Partly, but confuses traffic with interest | Answers it |
| Did people engage or just pass? | No | Answers it |
| Where did people give up? | No | Answers it |
| Did anyone come back? | No | Answers it |
| What is this space worth? | No | Answers it |
Scroll the table sideways to see all columns.
Four ratios turn presence into behavior, and none of them require knowing who anyone is:
These are comparable across spaces of different sizes and traffic levels, which raw counts never are. That comparability is what makes them usable for pricing decisions.
Counts describe scale. Ratios describe performance. Only the second can price a space.
An organizer selling sponsorship, a landlord setting rent by location, a retailer deciding where a category belongs. All three are effectively pricing attention, and all three usually do it with traffic numbers plus tradition. The moment you can show that zone A delivers three times the engaged time per square metre of zone B, the negotiation changes character. It stops being about position on a map and starts being about evidence.
Group-size bias. People arrive in groups. Raw presence counts reward spaces that attract pairs and trios over those attracting solo decision-makers, who are frequently the buyers. Normalise per capita or the numbers mislead consistently.
Adjacency contamination. Anything near an entrance, a toilet or a coffee point inherits traffic that has nothing to do with its own appeal. Either model the adjacency or compare only like-for-like positions. This single effect probably causes more bad layout decisions than any other.
You do not need to replace a counting system to start measuring behavior, and in most cases you should not. Counting handles safety and capacity perfectly well.
The upgrade is a layer, not a replacement: keep the counts for operations, add the four ratios for commercial decisions, and segment both by visitor category wherever your registration or ticketing data allows. The segmentation is usually where the first genuinely surprising insight appears. The group you designed the space for is often not the group that used it.
People counting measures how many individuals are present in a space over time. Behavior analytics measures what those people did: whether they approached, engaged, bounced or returned, and how that differs by zone and visitor type. Counting tells you how busy a space was; behavior analytics tells you whether it worked.
Not on its own. Dwell time mixes genuine interest with queuing, waiting and navigational confusion, which push commercial value in opposite directions. Dwell becomes meaningful only when paired with context such as attraction rate, depth into a space and whether the visitor returned later.
Attraction rate is the share of people passing a space who turn toward it and approach. It separates appeal from raw footfall, which matters because high-traffic locations near entrances often record large counts alongside poor attraction. It is one of the few metrics that compares fairly between spaces of different sizes.
Yes. Attraction, engagement, bounce and return rates are all computed from anonymous movement and require no personal data. Identity becomes relevant only when the goal shifts from understanding a space to generating named leads, which is a separate consent tier.
Visitors frequently arrive in pairs or groups, so raw counts overstate the performance of spaces that attract social browsing and understate those attracting solo decision-makers, who are often the actual buyers. Normalising per capita corrects this; without it, layout and pricing decisions are made on systematically skewed data.
