 |
Overview |
| SenSource provides the following information as a general guideline for the listed topics. Our hardware and software provides the tools necessary to collect customer traffic and provide related business performance metrics. However, since each application may vary and require special needs and considerations to achieve the desired goals, we are available for consultation, special programming, and/or database integration as needed. |
| |
|
Content |
|
| |
|
Information |
» Acquiring Accurate Traffic Data
- Counters always work best when they are monitoring traffic in a area where people are in motion (either walking IN or OUT) below the sensor.
- Customers or employees who congregate or loiter beneath the sensors may create errors in the counter values.
- A floor plan conducive to creating a good traffic flow below the sensor will produce positive results.
- When a sensor is mounted near an entrance or exit with immediate exposure to outside weather conditions (especially in areas with extreme weather variations), the OUT count is the most accurate value to use in collecting accurate customer traffic data.
- Another method commonly used is to take the average of the two counts:
» Certifying or Validating Counter Data
- The most reliable method to validate the accuracy of a particular system is to compare hand-counted customer traffic (using reliable staff) to data collected by the SenSource people counting system over the same period of time.
- This is the method used during the original installation.
- The most commonly used method (post installation) to use during normal operations is to compare total daily IN counts to OUT counts using the following calculation:
- [(OUT - IN) ÷ OUT] ÷ 2 = ΔV% (Variation)
- Note: This calculation assumes that neither the IN count or the OUT count is 100% accurate. It also assumes a best case scenario that the actual count is in the middle of the IN and OUT counts.
- If the percent variation exceeds +/- 10% for seven consecutive days, this is a good indicator that there is a problem at a particular facility, and further investigation may be required.
- Thermal sensors, that are mounted near an area exposed to extreme variations in weather conditions, have exhibited anomalies.
- Other external factors that may create errors in traffic counts are significant changes to traffic patterns causing customers or employees to congregate or loiter below the sensing area.
- Improper installation may also contribute to errors in accuracy.
» Cleansing or Normalizing Data
- When data anomalies are detected, a "cleansing" or "normalization" routine may be applied.
- Various methods have been incorporated to accomplish this based on the acceptable models for each application. Some of the most common methods are as follows:
- Normalizing - Replace erroneous or missing data for a particular day with the average for the previous four same day of the week (i.e. if Wednesday’s counts are bad, then replace it with the average of the previous 4 Wednesdays).
- Smoothing - Replace erroneous or missing data with the average of the previous day and the following day. (i.e. if Wednesday’s counts are bad, then replace it with the average of the previous Tuesday and the following Thursday).
- Cleansing - Remove sporadic daily data from calculations used to determine weekly, monthly, or yearly statistics and averages.
» Calculating Conversion Ratio
- The conversion ratio depicts what percentage of customer traffic converts to a sale.
- Determining the conversion ratio is a key factor in determining staff levels and staff quality based on key performance metrics of a particular facility, region, or group of facilities. The basic method used to calculate the conversion ratio for a given time frame is as follows:
- Total Transactions (TT) ÷ Total Customers (TC) = Conversion Ratio (CR)%
- Total Sales Volume (SV) ÷ Total Transactions (TT) = AVG Sale
- Total Sales Volume (SV) ÷ Total Customers (TC) = AVG Sale Distribution
- Total Sales Rep Transactions (TSR) ÷ Total Customers (TC) = Sales Rep Conv. Ratio
- Other vital statistics can be calculated as well.
|
|
