How to Count Window-Front Traffic: A Measurement-Line Acceptance Sheet
Define pass-by lines, entrance coverage, queue rules, and matched observation periods before reporting a storefront ratio. Includes a counting acceptance sheet.

Consider a corner site with two entrances, a queue that doubles back past the window, and a service door that customers use at lunch. A count designed for one straight frontage will produce a number, and the number will not survive the first review. The useful work is not the tally. It is the definition of the lines, directions and exclusions that the tally follows.
This is a method for measuring pedestrian traffic outside a window display. It is not a report template, and it does not turn the display into a measuring device. The reporting rules and the evidence-grade questions belong with the signage analytics guide and the proof-of-play guide. This sheet is the part that comes before either: whether the measurement definition is good enough to collect anything worth reporting.

What the sheet records
Passing traffic is a count of crossings at a defined line in front of the frontage. Entries are crossings at the entrance line or lines that serve that frontage. Both are crossing events, not a count of unique people, and neither one implies the other. The sheet exists so that a second person can reproduce the count without renegotiating what it meant.
| Field | What must be written |
|---|---|
| Frontage boundary | The measured frontage, drawn on a plan, with the neighboring frontage excluded |
| Passage line | Exact location, endpoints, direction or directions counted, and any point where the view is obstructed |
| Approach routes | Every route by which a person can reach an entrance, including turns at the corner |
| Entry lines | Each entrance that serves the frontage, or an explicit statement that only some entrances are covered |
| Queue and return rules | How a queue, a U-turn or a re-entry is treated, and which observable crossing events qualify; do not infer unique people from aggregate totals |
| Group rule | How two people walking together are counted, applied consistently |
| Exclusions | Staff, deliveries, contractors, and anyone else not counted |
| Observation periods | Dates, start and end times, and the trading conditions represented |
| Observer positions | Where each observer stands, and how their totals are synchronized and reconciled |
| Sensor boundary | If a sensor is used, the line it actually covers, how it was validated, and where it differs from the manual definition |
| Missing data | Which intervals are incomplete, and the rule for excluding the same interval from both counts |
Three failures, and the definition that fixes them
The corner route never crosses the line. A passage line drawn across the shopfront may miss the approach that turns into the entrance from the side street, or may be crossed twice by the same person. The fix is a route diagram, not a longer line: mark every approach, place the passage line where the approaches actually pass, count directions separately, and state whether a turn into the entrance counts as a passage. If an approach cannot be captured, write that down and limit the claim.
The queue is counted as traffic. A person who joins a queue inside the window zone may cross the passage line, stop, shuffle forward and cross again. Without a rule, the same visit becomes several events. Define qualifying crossings and record observed queue movements separately. Keep interval-level queue notes and exclusions. Timestamps on aggregate totals cannot establish that two crossings were the same person; do not claim unique-person de-duplication without a separately validated method.
Only one of two entrances is covered. The second entrance affects the entries count, which is the numerator of the ratio, not the denominator. Leaving an eligible entrance uncounted understates recorded entries. With the same valid passage denominator, it understates the ratio. If frontage coverage also differs, the direction of bias is not established. Label partial entrance coverage and do not present it as an all-site capture rate. The alternatives are to cover every entrance, to sample an entrance and report it as a sample, or to refuse the ratio until the coverage exists.
Keep the two streams separate, and keep missing data honest
Count passages and entries independently, then reconcile the interval totals. Do not adjust one stream to fit the other. Preserve the raw counts in each stream. If one stream is missing for part of the day, use only matched complete intervals in the ratio calculation and flag the excluded intervals; a missing count is not zero traffic, and a zero denominator has no defined ratio. Label the coverage rather than extrapolating it.
If a sensor is proposed, treat it as a separate instrument with its own boundary. Validate it against a manual reference in the same interval and at the actual placement, including the light conditions that change during the day, and validate a through-glass placement specifically if that is what is proposed. Report the signed count difference; calculate percentage error only when the manual reference is nonzero, and keep the count difference visible. Align timestamps before comparing anything. A sensor that covers a different line is a different measurement, and merging the two produces a number that belongs to neither.
The arithmetic, and its limits
Passage and entry counts produce a ratio of entries to passages for the counted frontage and period. Suppose a hypothetical frontage records 640 passages and 96 entries in the same trading period. The ratio is 96 ÷ 640 × 100, or 15.0%. If the same period also records 12 purchases, then 12 ÷ 96 is 12.5%, but that is purchases per entry, not a conversion rate, unless entries have been defined as visits and repeated crossings have been handled. Two ratios, two definitions, two owners.
A zero denominator has no defined ratio. Show the counts alongside the ratios. Small samples need uncertainty acknowledged and do not support firm claims from a percentage-point difference alone. Do not import a benchmark from another site, and do not treat a ratio from a covered entrance as if it described the whole storefront.

What the display does and does not do
This display does not establish a counting capability, and it does not identify anyone. If measurement is part of the project, it is a separate scope with its own method, coverage and responsible person. Aggregate counts only; no demographic or attention outputs are implied. If cameras or sensors are proposed, notices, retention, permissions and any legal review belong to the site owner and its advisers. A method that does not identify people is not automatically outside privacy rules.
For reporting, the analytics guide covers how delivery, exposure and response should be described. This sheet only makes sure the pavement numbers underneath those descriptions have a defensible definition.
Use it in a pilot
The sheet is most useful in a pilot where the question is whether a window display changes the street-to-entry ratio. Pick sites by the conditions that change the viewing result, as the rollout pilot guide describes, and alternate the proposed window content with the incumbent over matched periods. Keep the frontage, the lines, the hours and the entry coverage identical across periods. Matched periods alone do not establish causation: weather, promotions, opening patterns and other changes can still confound the comparison. Keep visibility acceptance separate from trading observation; the retail placement guide covers the position and sightline work that comes before counting.

Send the measurement plan with the display brief
Bring the marked frontage, the entrance plan, the observation periods and the question you want the count to answer. We can configure the screen for the position you intend to measure and be clear about what the hardware does not do. Request a Quote, and see the Slim Bezel High-Brightness Window Displays in the window displays collection.
Questions the store team asks
Does a camera count prove people looked at the window? A crossing count alone does not establish attention or recall. Any separately proposed attention measurement needs its own validated definition, method and privacy review.
Is the entry-to-passage ratio a conversion rate? No. It is entries over passages for a covered frontage. Conversion needs visits and purchases defined for the same period.
Can we compare two stores? Only if the frontage coverage, hours, line definitions and entrance coverage are comparable, and even then as an operational comparison rather than a causal result.
What if the second entrance is missed? The entries count is incomplete. With an unchanged valid passage denominator, the ratio is understated; if coverage differs in both streams, its bias is not established. Either cover it, sample it and label the sample, or do not report a capture-style ratio.
Is a count with no personal data outside privacy rules? Not automatically. Aggregate counts reduce exposure, but notices, retention and permissions remain decisions for the site owner and its advisers.
Sources and scope
- V-Count, Storefront Counting & Capture Rate — definitions of passing traffic and entries, the entries-over-passages ratio, missing-denominator handling, validation against a manual reference, and the caution that aggregate stages are not the same people. Used as a method reference, not an industry standard.
- RetailNext, Passby Traffic and Capture Rate — the passby and capture definitions and the window-display use case. A vendor brief, not an independent benchmark.
- The numerical example is hypothetical. No display in this article counts people, identifies people or measures attention, and no volume, benchmark or causal effect is claimed.
Request a Quote

