retail analytics

Queue management that warns before the wait becomes a complaint

In short

Queue management applies Ayonix dwell measurement and people counting to a defined queue area. It reports how many people are waiting and how long they have been waiting, and raises a warning when either exceeds the service standard, so a supervisor can open another position before customers abandon the queue.

Request a pilot Send a sample video

Conceptual illustration of checkout queueing used for wait measurement. AI-generated conceptual image, not a customer deployment.

Scope

What this analytic does, and what it does not

What it detects

Number of people inside a defined queue area
Time individuals have been waiting in that area
Queue length and wait time exceeding configured thresholds
Rate of growth, distinguishing a building queue from a stable one

What it does not guarantee

Exact wait time for any specific customer
Reliable measurement in an unstructured queue with no defined area
Distinction between a queue and people standing nearby for other reasons
Any prediction of how long a queue will take to clear

The problem

What customers are actually dealing with

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

Supervisors find out about a queue when a customer complains or abandons their basket.
Additional positions are opened too late because nobody is watching the queue continuously.
Service-level targets exist on paper but are never measured, so nobody knows whether they are met.
Staffing at peak is argued from anecdote because there is no wait-time record.

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

Queue-area cameras are read and people in the defined area are detected and tracked.
2

Measure

Queue population and per-track waiting time inside the area are accumulated continuously.
3

Evaluate thresholds

Queue length and wait time are tested against the configured service standard.
4

Stabilise

A sustained-duration requirement prevents a momentary bunching from raising a warning.
5

Deliver

Warnings route to the supervisor channel, and the wait series is written to the dashboard.

Configuration options

Queue area polygon per lane or per queue group
Wait-time and queue-length thresholds reflecting the service standard
Sustained duration before a warning is raised
Trading-hours schedule and different thresholds by period
Alert routing to handheld devices, dashboards or the VMS
Grouping of lanes so a group warning is raised rather than one per lane

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.

A sustained-duration requirement prevents a brief bunch of arrivals from triggering a warning.
Grouping lanes avoids six simultaneous alerts for one queueing situation.
A minimum dwell before a person is counted as queueing excludes people passing through the area.
Cooldown prevents repeat warnings while the same queue condition persists.
Measurement degrades in unstructured queues; the pilot establishes whether the queue area is well enough defined.

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

A view covering the full extent of the queue at its longest, not just the head of the queue
An elevated angle so people standing behind each other remain separable
Consistent lighting across the queue area through all trading hours
Person height of at least 8–10% of frame at the tail of the queue
Stable mounting so the queue polygon remains aligned

Environmental limitations

A queue that extends beyond the camera view will be under-reported at exactly the moment it matters.
Unstructured or branching queues are much harder to measure than a defined lane.
People standing in the area for other reasons inflate queue population.
Trolleys and spacing behaviour differ by store format and affect both length and wait measurement.

Delivery

Alerts, evidence and where they land

Event and evidence fields

Queue area or group name
Queue population at trigger
Median and maximum wait at trigger
Threshold crossed and duration above it
Camera and site
Timestamp with timezone
Event identifier

VMS integration

Queue warnings are usually delivered to supervisor devices rather than the VMS alarm queue, because the responder is on the shop floor rather than in a control room. Where a control room does monitor queues, a VMS custom event with the queue camera is appropriate.

See compatibility states →

Deployment options

Edge processing per store so queue video stays on site
On-premise processing where checkout cameras already aggregate locally
Multi-site deployment so the same service standard is measured consistently across the estate

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

Median and 90th-percentile wait per queue group per interval
Queue population trend
Threshold breaches per day and per hour of day
Time to open an additional position after a warning
Percentage of trading hours within the service standard
Camera availability per queue area

Measurable pilot criteria

Wait-time accuracy against stopwatch ground truth for sampled customers
Queue population accuracy against a manual count at sampled times
Warning latency from the true threshold breach
False warnings per trading day from people passing through the area
Behaviour when the queue extends to its maximum realistic length

Governance

Privacy and governance

Wait time is measured per anonymous track and aggregated before reporting.
No identity is established and no profile is retained across visits.
Where queue areas include staff positions, employee monitoring obligations may apply.
Aggregate wait series can be retained without imagery.

Frequently asked questions

Can it measure an unstructured queue?

Less reliably. A defined lane or a marked waiting area produces good measurement. A queue that forms in an open concourse with no fixed shape is much harder, and the readiness review will say so rather than promising a figure that will not hold.

Is the wait time exact for each customer?

No. It is a measurement of time spent in the queue area by an anonymous track, and it can be affected by occlusion and track breaks. Median and percentile figures across many customers are far more dependable than any single reported wait.

Does this need new cameras above every till?

Usually not. One elevated camera covering a group of lanes is often better than several low cameras, because it sees the full queue extent. Camera suitability is assessed per queue area during the readiness review.

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.