5 Reasons Your Wi-Fi Dwell Times Are Irrelevant

Technologies to measure it abound.  Wi-fi technology has a natural appeal for delivering shopper dwell time because it is increasingly required for a multitude of reasons.  The same technology platform justified by store associate productivity initiatives such as handheld devices can also be used with little or no extra cost for shopper insight.

As wi-fi technology evolved engineers began to expose the simple device detection data that each access point naturally observes.  That data can be processed to determine timeframe over which a device was observed by an access point or network of access points. 

Most wireless hardware and wi-fi analytics companies call this dwell time.  It is a great name from a marketing standpoint because dwell time is a highly significant statistic in retail.

Unfortunately wi-fi vendor "dwell time" is not retail dwell time.   In fact out of the box wi-fi vendor dwell time is irrelevant to retailers.  We recently compared visitor counts from a wireless product dashboard against actual visitor counts and the wireless dashboard was over 5 times higher than actual.

Wi-fi shopper data tends to have one or more of the following five problems, each of which make it all but useless as a retail metric.  Wi-fi vendor "dwell time" is more accurately device detection duration. 

All 5 problems discussed here must be resolved in order to compute retail dwell time

1. Includes dwell time outside work hours

Most dwell time calculations we see are counting device detection time 24 hours a day 365 days a year.  For many stores, the closed store device detection window is larger than the open store window.  Aggregations of dwell time are inaccurate including daily dwell times or average dwell times per store.

2. Includes employees, wi-fi enabled equipment and wi-fi enabled products

A wall full of wi-fi enabled televisions running 24 hours a day can significantly inflate dwell time.  It's the same for employees.  When shopper dwell times are typically one hour or less, the detection of wi-fi enabled store equipment and every individual powered-on product from thermostats to refrigerators to watches makes wi-fi dwell time irrelevant.  A recent analysis of visit duration metrics with employees and static devices and then without them showed that average dwell times across 1000's of shopper visits were twice what they were without employees and static devices.  The stock dwell times were completely wrong.

3. Ambiguous on whether dwell time is in store or not

Wi-fi signals do not stop at the boundaries of the store interior.  The store real estate impacts the interpretation of the numbers.  Free standing stores will have less concern on accuracy but mall or strip shopping center locations will capture more noise.  Vendor wi-fi dwell time is simply devices detected.  The accuracy of location sensitivity is a function of the wi-fi architecture, it's configuration and it's physical implementation along with processing logic that understands the data.  None of this is accounted for and most of it is not configurable in the product demonstration the vendor just showed you.

4. Doesn’t connect visits across cell phone disconnect for single visit

The wi-fi signal is not an always-on signal.  Therefore, a definition of a visit and a detection duration is not a single connection duration, it is a computed duration based on an algorithm that puts together discrete signals from a single device.  The out-of-the-box logic will have difficulty differentiating between a device that was present and have gone inactive but is still present, from a device detected by a passerby who passed twice within 20 minutes.  In both cases the signal was inactive for a period and then became active so the wi-fi product uses a single algorithm to classify both cases the same.  Because any significant sophistication in the algorithm requires contextual knowledge of the store and access points, the logic tends to inflate visits and decrease average dwell time.

5. Handling of randomized mac addresses

Several phone operating systems, notably the latest Apple IOS's, publish obscured identities when probing access points until the user associates to the network, which releases the actual identity.  This is called MAC address randomization.  When an access point detects a device it gets a MAC address which may be actual or not.  When it detects the same device with a randomized MAC address a second time it gets a different MAC address.  Therefore, the device (MAC address) count is incremented every time the access point observes the unassociated device.  The dwell time is either zero minutes or one minute.  Many wi-fi access points pass on the randomized MAC address without filtering.  This can significantly impact dwell time computations.

"Half of knowledge is knowing where to find it"

Translating vendor device detection dwell times into usable, valuable retail dwell times can be accomplished through minimal investment.  If you want to detect retail dwell times using the wi-fi architecture you've already purchased and implemented then you would benefit from the experience of a consulting firm that understands retail and wi-fi technology.  Our customers are seeing immediate ROI from short consulting projects.

For information on how to get started, schedule an intro call with Quality Deployment Managing Director, Michael Lee.