A footfall number is useful only when everyone understands what it counts. Entrance passages, people currently inside and unique visitors are different measures. Calling all three “customers” can make a report look precise while answering the wrong question.
Before evaluating CCTV people counting, define the business decision the number will support. Staffing a busy entrance, comparing trading periods and reviewing occupancy need different scope and validation.
Write the counting definition
Record the entrance or boundary, directions, operating hours and reporting interval. Decide how staff movements, repeat entries and deliveries should be handled. Ask the supplier which of those distinctions the proposed system can actually make.
For a multi-entrance building, identify every relevant entry and exit. Establish whether the proposed report combines those streams and what happens when one is unavailable. A total from one door should not be labelled whole-site footfall unless the scope supports that description.
Do not assume that a counting system identifies individual people. Where unique visitor analysis is proposed, ask what method and data it requires and assess that separately from simple passage counts.
Check the entrance against product requirements
A camera installed for general recording may not provide the view needed for counting. Ask the supplier to assess mounting position, coverage, lighting, obstructions and the behaviour of people at that entrance.
Axis's counting guidance emphasises suitable installation conditions and comparison with manual observations. It also distinguishes different counting solutions rather than treating one method as appropriate everywhere. Read the manufacturer's counting guidance.
Use the requirements for the actual product under evaluation. Do not copy another manufacturer's mounting recommendation into a specification for an untested system.
Compare against an independent sample
Agree the sample plan before examining the automated results. Have an authorised observer count the same defined passages independently during the selected periods. Match the start and end times and record the relevant direction.
Include periods that represent the business: quiet entries, busy groups, staff activity and any conditions that make the doorway difficult to see. Record what was excluded, such as a temporary closure or an unavailable camera.
Compare the counts by period and condition. A single overall total can hide overcounting at one time and undercounting at another. Keep the sample size visible and avoid describing a short successful test as permanent accuracy.
Decide what the report may be used for
Set acceptance criteria appropriate to the intended business decision. A trend report and a process requiring an exact reconciled headcount have different needs. Do not quietly change the claimed purpose after seeing the result.
If you compare footfall with sales, align the store, dates and trading hours. Explain the numerator and denominator: a transaction count is not automatically a count of different purchasing people, and an entrance passage is not automatically a unique shopper.
Keep unavailable periods and material configuration changes visible. A decline caused by a missing camera interval should not be reported as a real decline in customer activity.
Maintain the definition as the shop changes
Recheck the installation after displays, door arrangements, lighting or routes change. Record software and configuration updates that could affect the count. A comparison between branches is useful only when their definitions and limitations are understood.
Assign a person to review exceptions and keep the report's scope attached to exports. This prevents a carefully qualified pilot number from becoming an unsupported claim when it is copied into a management presentation.
AI Alerto by PGAK can review your entrance layout and the measure you need. Prepare the number of entrances, existing camera details and the intended reporting use. The video analytics guide explains the broader buying questions, while the retail evaluation guide helps define a manageable first project. Compatibility and the final counting method must be verified for the proposed setup.
Common questions
Is a footfall count the same as unique visitors?
Not necessarily. Define whether the system measures passages, direction, occupancy or unique people. Repeat entries and several entrances can change the interpretation.
Can counting errors be assumed to cancel out over a week?
No. Errors can be systematic. Validate representative periods and review changes in the scene or configuration before treating a longer reporting window as reliable.