Measuring counter service time without monitoring employees

Measure the process, not the person

In a quick-service restaurant, a café, a bakery or a fuel station shop, counter service time has a direct impact on the customer experience. Long waits at peak times drive customers away; a poorly calibrated set-up wears teams out. Yet this time is rarely measured objectively: managers rely on impressions or the odd bit of feedback.

Video can provide that measurement, on one essential condition: what is being measured is the service process, not the person serving. The aim is not to know how long a particular employee takes to prepare an order, but how long customers wait at the counter on average, at what time and under what conditions.

This distinction is not just a matter of principle. A system designed to assess individual employees would raise issues under Law 09-08 and employment law, and would erode trust within the team. VIDEO.MA does not offer individual employee monitoring: our analysis covers zones and aggregated indicators.

Defining where a service cycle starts and ends

Before installing anything, everyone needs to agree on what is being measured. "Service time" can mean several different things, and each translates into a different configuration:

• queue waiting time, from entering the queue zone to reaching the counter;
• time at the counter, from the customer arriving at the till to leaving the zone;
• total time, from joining the queue to leaving the counter or the pick-up area;
• for formats with a separate pick-up point, time spent in the order pick-up zone.

Events observed on the customer side

In every case, the start and end of the cycle are defined by events observed on the customer side: entering and leaving a zone drawn on the image. This is deliberate: it measures service without depending on what staff do or where they move.

Edge cases need planning too: a customer who leaves the queue without being served, a group ordering together, someone waiting for a friend without ordering. Not all of these are handled automatically; they are documented, and their effect on durations is checked during the pilot against manual observations. Results are best expressed as a median and a distribution rather than a simple average, which a few extreme cases can skew.

Framing focused on the customer area

Respect for the team is largely decided at the framing stage. A camera behind the counter pointing at staff is not suitable for this approach, even if it already exists and would be convenient to reuse. If no camera covers the queue and the front of the counter from the customer side, adding a dedicated, properly aimed camera is usually the better option. We favour cameras aimed at the customer area, with settings that limit what is analysed:

• analysis zones limited to the queue and the front of the counter;
• privacy masks over the work area behind the counter, the kitchen and preparation stations;
• no analysis of workstations, breaks or staff movements;
• on-site processing (Edge AI), producing durations and counts without moving footage around;
• no facial recognition or identification of anyone, whether customers or staff.

Using the data: peak hours, organisation, staffing

Once durations have been measured and validated, their value comes from aggregating them over time. A useful dashboard shows median waiting time by time slot, observed queue length and how both change across the days of the week.

This data supports operational decisions: adding staff when waits get longer, opening a second till at the right moment, preparing ahead of a recurring peak, or assessing the impact of a counter layout change. Comparing several outlets measured the same way also helps spot differences in how sites are run.

Linking these durations to order numbers or revenue requires integrating point-of-sale (POS) data, which depends on its export or API capabilities and calls for a specific study. In the dashboard, what comes from video (durations, queues) must stay clearly separate from what comes from the till (receipts, amounts). And these indicators remain collective: they are not designed to evaluate an employee and must not be repurposed to do so.

Informing and involving your teams

A service time project works better when it is explained. Before going live, the business, as data controller, informs staff about the cameras, the purpose of the analysis, the data produced and who can see it, and consults staff representatives where required. Customers are informed through appropriate signage, and the applicable formalities are completed with the CNDP.

In practice, a short briefing note can answer the questions staff are most likely to ask: what the cameras analyse, what they do not analyse, what the dashboard shows, who can see it, how long data is kept and who to talk to with questions or concerns. Showing the team an actual dashboard screen, with its aggregated figures and no images, is often more reassuring than any explanation.

Sharing the results with the team is often the best way to embed the approach: they are the ones handling the rush, and the data gives them evidence to ask for extra help or a better set-up. To learn more, see our Restaurants & cafés and Fuel stations pages, or get in touch to scope a pilot at one of your outlets.