How to Measure a Taxi-Screen Campaign Without Confusing Plays and People
A play log tells you an advert ran. It takes different evidence to show that a person saw it or became a customer.
Plan your campaign
Outdoor campaign dashboards can show several numbers that look similar but answer different questions. Before judging a campaign, separate delivery, audience signals and business results. This guide explains how to interpret those layers when reviewing a taxi-screen campaign with Active Vision Media.
Start with evidence that the advert played
Proof-of-play is a record of an advert being played on a device. Useful fields include the creative identifier, screen identifier, timestamp and playback duration. A GPS record can help place the vehicle at the time of delivery when a recent location is available. Check that the creative identifier belongs to your advert and the screens belong to your campaign. A scheduled slot is not the same as a confirmed play. Likewise, the total number of ads a screen played is not the number of times your own ad played. Reporting should filter by your campaign’s actual media and device assignments. A screenshot is a spot check at a moment in time; it cannot replace the history of delivery.
Read share of voice with its time window
Share of voice is your allocation within an advertising loop or defined period. A 25% allocation means roughly one quarter of the planned advertising time in that loop; it is not a promise that one quarter of people nearby looked at the screen. Different creative lengths, offline periods and schedule changes can affect the number of plays delivered. Compare the allocation with actual delivery for the same dates.
What detected footfall can and cannot tell you
A detected-footfall count is an audience signal, not a receipt for human attention. Where counts come from nearby wireless-device detections, devices and people are different units. One person can have multiple devices; others may not be detected; a device may be seen repeatedly. Traffic, distance, sensor settings and deduplication rules can change the count. Ask which sensor produced the count, what area and time period it covers, how repeated detections are treated and when the data last refreshed. A sensor count near a vehicle should not automatically be attributed to every advertiser on that vehicle. Linking a count to your advert requires a documented time-and-location method. Neither a nearby device nor a passing pedestrian proves that someone read the advert.
Separate measured figures from estimates
Confirmed plays: recorded delivery events for your own creative. GPS and online status: the most recently received device information, with an age or timestamp. Detected footfall: the sensor’s recorded proximity signal, subject to its methodology. Estimated impressions or reach: a modelled audience figure with stated assumptions. Enquiries and sales: customer actions recorded through your own business systems.
Connect delivery to a business outcome
Give customers a simple response path: a memorable URL, a campaign-specific offer or a clear booking instruction. Ask new enquiries how they heard about you, and retain the answer with their order if appropriate. If you use QR codes, design them for safe pedestrian or passenger use rather than encouraging drivers to interact with a phone. Record the campaign spend, attributable enquiries and completed purchases over a stated period. Cost per enquiry is spend divided by qualifying enquiries; cost per acquired customer is spend divided by attributed new customers. If there are no attributed customers, report that result plainly. Do not divide by an invented conversion count or substitute footfall for sales.
Make your end-of-campaign review useful
Compare delivered airtime with the agreed booking first. Then review which creative, dates and locations coincided with relevant responses. Missing sensor readings should be identified as missing, rather than silently converted to zero. Your conclusion should say what was measured, what was estimated and what remains uncertain. That makes the next budget decision more useful than a large headline number alone.
Are detected footfall and verified viewers the same?
No. Proximity signals, particularly wireless-device observations, do not prove unique people or that someone saw an advert.
Does 25% share of voice mean my ad gets 25% of viewers?
No. Share of voice describes an airtime allocation within a defined loop or period, not the share of people who saw the advert.