retail analytics

Heatmap analytics that show which parts of a space are actually used

In short

Ayonix heatmap analysis visualises the areas of a camera view that attract the most activity, accumulated over a chosen period. It shows which aisles, entrances and zones are used and which are ignored, providing an aggregate basis for layout, staffing and merchandising decisions without identifying anyone.

Request a pilot Send a sample video

Conceptual illustration of shop-floor movement used for heatmap analysis. AI-generated conceptual image, not a customer deployment.

Scope

What this analytic does, and what it does not

What it detects

Accumulated presence intensity across the camera view
Relative activity between areas of the same floor
Change in activity distribution between comparable periods
Areas with consistently low activity across the reporting period

What it does not guarantee

An exact number of people in any area — a heatmap shows relative intensity
Comparison between cameras with different angles or coverage
Any causal explanation for why an area is busy or quiet
Individual movement paths or identification

The problem

What customers are actually dealing with

These are the situations that lead teams to look at this analytic in the first place.

Floor space is allocated by convention and nobody knows which areas are genuinely dead.
A refit is proposed with no measurement of how the current layout is used.
Staff report that an aisle is quiet but there is no evidence to support reallocating space.
Seasonal changes in how a space is used are invisible because nothing is measured continuously.

How it works

Detection workflow

Every event carries the rule that produced it, so an operator can see why they were alerted.

1

Ingest and detect

Floor cameras are read and people are detected and tracked.
2

Project to ground plane

Track positions are mapped to floor positions where the camera calibration supports it.
3

Accumulate

Presence is accumulated into a spatial grid over the reporting period.
4

Normalise

The grid is normalised so intensity is comparable across the period and against a baseline.
5

Publish

The heatmap is rendered as an overlay with a legend and stored for period comparison.

Configuration options

Accumulation period — hourly, daily, weekly or a custom campaign window
Grid resolution, balancing spatial detail against statistical stability
Areas of the frame excluded from accumulation, such as a window or a service corridor
Normalisation basis so periods of different length can be compared fairly
Trading-hours schedule to exclude restocking and cleaning activity
Baseline periods retained for before-and-after comparison

Alert quality

How irrelevant and duplicate alerts are reduced

No video analytic eliminates false alerts. These are the mechanisms that reduce them, and the residual rate is measured on your own cameras.

Longer accumulation periods produce more stable maps; a single hour of data is rarely conclusive.
Excluding closed hours prevents staff activity from dominating the map.
Normalisation makes periods of unequal length comparable rather than merely adjacent.
Grid resolution is chosen so each cell accumulates enough observations to be meaningful.
A heatmap shows relative intensity within one camera view and should not be read as a count.
Accumulated intensity still carries perspective bias and occlusion artefacts that no accumulation period corrects. The map is indicative of relative activity rather than a measured count, and comparisons between cameras remain invalid unless the views are genuinely equivalent.

Prerequisites

Camera requirements and environmental limits

Camera suitability decides more of the outcome than any software setting. These are assessed per camera before commitment.

Camera requirements

An elevated, downward view over the floor area — the higher and more vertical, the better the spatial mapping
Coverage of a complete, meaningful area rather than a fragment of an aisle
Even lighting across the floor area, avoiding a mix of bright and deeply shaded regions
Stable mounting, since any camera movement invalidates accumulated spatial data
Sufficient resolution at the far end of the covered area

Environmental limitations

Perspective distortion means a near-horizontal camera produces a heatmap that over-weights the foreground.
Fixtures that occlude the floor create artificial cold spots.
Maps from different cameras are not directly comparable unless the views are genuinely equivalent.
Camera movement, even slight, breaks comparability with earlier periods.

Delivery

Alerts, evidence and where they land

Event and evidence fields

Camera and floor area
Accumulation period start and end
Normalisation basis
Grid resolution
Excluded regions
Baseline period reference where a comparison is stored

VMS integration

Heatmaps are a planning output rather than an operator notification and belong on the operational dashboard. They are typically exported for use alongside floor plans in merchandising and facilities reviews.

See compatibility states →

Deployment options

Edge processing so only the accumulated grid, not video, leaves the site
On-premise processing where floor cameras already aggregate locally
Multi-site deployment where comparable store formats are benchmarked against each other

Compare architectures →

Measurement

Dashboard metrics and pilot acceptance criteria

Acceptance thresholds are agreed with you before the pilot starts. This site publishes no benchmark figures, because they do not transfer between sites.

Dashboard metrics

Accumulated intensity map per camera per period
Ranked zones by intensity
Period-over-period change map
Percentage of floor area in the lowest intensity band
Campaign window comparison against baseline
Camera availability during accumulation

Measurable pilot criteria

Map stability between two comparable weeks under unchanged layout
Agreement between measured high-intensity areas and independent observation
Sensitivity check — does a deliberate layout change produce a visible, expected change?
Data completeness across trading hours for the accumulation period
Confirmation that excluded regions behave as configured

Governance

Privacy and governance

The output is an aggregate spatial grid. It contains no identity, face template or individual path.
Grids can be retained long term without retaining any imagery.
Where a view includes staff working positions, employee monitoring obligations may apply.
Grid resolution should be coarse enough that individual behaviour cannot be reconstructed.

Frequently asked questions

Is a heatmap a count of people?

No. It shows relative activity intensity accumulated across a period within a single camera view. For numbers, use people counting or occupancy. The two are complementary: counting tells you how many came in, the heatmap tells you where they went.

How long should we accumulate before drawing conclusions?

Long enough to cover the variation you care about. A single day reflects that day’s weather and promotions. Most layout decisions use at least two to four weeks so weekday and weekend patterns are both represented.

Can we compare heatmaps between two stores?

Only where the camera views are genuinely comparable in height, angle and coverage. Otherwise the differences in the map reflect the cameras rather than the shoppers. Cross-site comparison is designed during the readiness review, not assumed afterwards.

Keep reading

A controlled pilot establishes what this analytic actually does in your environment, against acceptance criteria we agree before it starts. Send a sample video first if you would rather see an assessment before committing to a pilot.