Part 1 — The Retail Challenge: Can Wi-Fi Tracking Deliver Actionable Precision?
Case study: Precision analysis and real-world value of Wi-Fi based geolocation with Cisco Meraki devices
Retailers are increasingly looking for ways to understand what’s happening inside their physical spaces with the same clarity they get online: which zones attract attention, where customers spend time, and how traffic flows through the store. That’s the promise of indoor location analytics—turning movement inside a venue into measurable insights you can act on.
In the broader indoor location world, you’ll often hear the term RTLS (Real-Time Location Systems). RTLS refers to systems designed to locate people or assets indoors, and Wi-Fi geolocation sits in the same family of ideas: devices are detected by multiple receivers, and the system estimates where they are on a map. The big difference is mostly about how the signal is collected and how frequently updates happen, not the goal itself. In retail, the goal is usually not to track someone “every second,” but to build reliable intelligence: heatmaps, dwell zones, zone counts, and customer journey patterns.
That brings us to the real question for stores:
Can Wi-Fi geolocation deliver enough precision to support decisions—not just pretty maps?
To answer it, our partner company Eye-In Media conducted a structured evaluation inside a large international sporting goods retail environment using Cisco Meraki access points. We measured how often location updates arrive, how stable the detected positions are during movement and dwell, where accuracy naturally improves or degrades, and what infrastructure changes can reduce errors—especially around the areas that matter most for retail analytics.
How Wi-Fi based geolocation works
When customers enter a venue with geolocation-enabled Wi-Fi infrastructure, their smartphones naturally emit and exchange Wi-Fi signals. Sometimes it’s because they’re connected to Wi-Fi, and sometimes it’s simply because the phone is scanning the environment. Access points can detect those signals, and the system estimates device location by comparing what multiple APs hear.
People often call this “triangulation.” The practical meaning is simple: the device needs to be heard by several APs at the same time so the system can place it confidently on a map. That’s one reason Wi-Fi location tends to work best in areas where AP coverage overlaps cleanly.
With Cisco Meraki, the location computations are handled on the backend side of the platform. Partners like Eye-In Media can then access aggregated location observations through the Meraki Scanning API / Location Analytics, typically received in periodic batches rather than as a continuous stream.
To learn more about Meraki Location Analytics and the related API:
Note about privacy: the Scanning API is designed with privacy safeguards, and Meraki provides mechanisms to anonymize identifiers so the data isn’t directly identified.
What is the value of geolocation data in retail?
Once you can estimate where devices are inside a venue, an entire layer of retail intelligence becomes possible.
You can generate heatmaps that highlight the most visited areas, compare traffic between departments, and identify which parts of the store naturally pull attention. You can also analyze customer journeys by aggregating coordinates over time—revealing common paths, loops, and friction points in the layout.
Example of customer data analyzed through Eye-In Customer Journey
Example of Heatmap that highlights the most popular areas
This data can also support geofencing experiences. For example, in a shopping mall where Wi-Fi is managed at the mall level, you can still personalize content based on the customer’s location—serving store-specific ads, promotions, or surveys depending on where the device is detected.
To learn more about Eye-In Heatmaps and Customer Journey capabilities: https://eye-in.com/wifi/analytics-heatmaps
Example of geofencing Wi-Fi Ads inside a shopping mall.
Data analysis approach
To get results that reflect real usage (not a one-off snapshot), the Eye-In team ran multiple tests on site and collected data over multiple days. We combined controlled movement tests (walking specific routes) with dwell tests (staying in place for several minutes), and we compared results with devices connected to Wi-Fi versus not connected.
Update frequency: how often do we receive location “pings”?
Meraki location observations are not continuous; they arrive periodically. Meraki generally describes an update cycle in the range of about a minute or two, and our real-world observations matched that behavior.
Across our dataset, we measured an average of 96 seconds between two location updates for the same device—comfortably inside the expected range.
In the above chart: the time between 2 pings for the same access point (in seconds)
This has a simple implication: Wi-Fi geolocation can be great for understanding where people are spending time and how traffic behaves by zone, but you should be careful when trying to reconstruct an exact second-by-second path in a medium-sized store. A customer walking at a normal pace can cross large distances between two updates. You can interpolate movement between points to make the journey look smoother, but the exact path taken between two pings may remain uncertain—especially if the customer walks quickly.
That said, for larger venues such as malls or airports, this cadence often becomes less of a limitation because zones are larger and movement between zones typically takes longer.
Precision: what the system reports vs what we observed
For each coordinate, the system provides an estimated precision in meters (an uncertainty radius). In our dataset, the average reported estimate was around 14 meters (46 ft).
At first glance, this can look high. But what we observed on site is that real behavior can be noticeably better than the estimate suggests, especially in areas where the device is well “surrounded” by AP coverage. The estimate tends to be intentionally conservative—essentially a safety buffer to avoid overconfidence.
In the above chart: the estimated precision of all coordinates received (in meters)
What happens when someone is moving?
During walking tests, results were surprisingly good in the center of the store, especially when the device was connected to Wi-Fi. In those conditions, the detected position aligned with the real location most of the time—good enough to understand what department we were in, what zone we were moving through, and which areas were being visited.
Chart above: comparison between the position detected by Meraki
What happens when someone is stationary?
When we stayed idle for a few minutes, the location points tended to tighten. In our tests, we saw cases where the effective behavior was closer to ~8 meters (26 ft) while walking slowly, and as low as ~5 meters (16 ft) when we remained in place.
This matters because many of the most valuable retail insights are not about “a perfect path every second”—they are about where customers stop, dwell, and spend time.
Again, with another path taken the system couldn’t follow us near certain APs.
Chat above: projection of the imprecision of locations drawn with circles, to see where the system would never have positioned us even taking the estimated precision into account.
Connected vs not connected: why it matters
One clear pattern was that connected devices tended to behave better. The simplest explanation is consistency: when the device interacts more regularly with the network, it’s more likely to be detected in a stable way by multiple APs over time. That stability helps the system estimate location more confidently and reduces jumpy behavior.
This doesn’t mean non-connected devices are useless. They can still be detected and still contribute to aggregate analytics. But if your goal is the best possible location stability, connected devices generally provide cleaner signals for the system to work with.
Dead zones: what they are, why they happen, and how they distort analytics
During the evaluation, we identified what we call dead zones—areas where the system repeatedly struggled to place the device correctly, even when we intentionally spent time in those zones.
A dead zone does not necessarily mean “no Wi-Fi.” In fact, the device can have a good Wi-Fi experience and still be hard to locate. The issue is not connectivity; it’s geometry. For Wi-Fi positioning to work well, the device needs to be heard by multiple APs at the same time, from different positions. Near edges and corners, the device is less “surrounded,” so the system has fewer strong reference points to compute a reliable location.
This is exactly what we observed: near corners and perimeter walls, the system often struggled. Instead of placing us near the edge, it would sometimes “pull” the location inward—often toward the middle of the store, where AP overlap is naturally stronger.
Why dead zones matter
Dead zones don’t just create messy dots on a map—they can create misleading business conclusions.
If customers regularly visit a perimeter aisle but the system repeatedly places them closer to the interior, you may wrongly conclude that a shelf is ignored, a corner is underperforming, or a display isn’t getting attention. In reality, traffic exists—but the location engine couldn’t confidently represent it in that part of the store.
In other words, dead zones are not a “minor technical detail.” They directly affect the reliability of zone analytics, aisle performance, and merchandising decisions.
What we learned
This evaluation confirmed something important: Wi-Fi geolocation can absolutely produce actionable retail analytics, even when the AP network was not originally designed for geolocation. In the central areas of the store, results were strong enough to support meaningful heatmaps, zone trends, and customer journey analysis.
At the same time, we also confirmed the main limitation: perimeter and corner zones are where accuracy tends to degrade, and that degradation can create dead zones that distort the story the data tells.
Three takeaways to remember
-
Wi-Fi geolocation is excellent for zone-level retail intelligence and dwell-based insights.
-
Update frequency (about every minute or two) is enough for analytics, but not for true real-time path tracking.
- If AP placement isn’t optimized for geolocation, edges and corners can produce dead zones that shift traffic into the wrong areas.
Conclusion: where this goes next
So, can Wi-Fi tracking deliver actionable precision? Yes—when the goal is realistic and the data is interpreted with the right expectations. Even without geolocation-focused optimization, the system showed strong value in large areas of the store and produced insights that retailers can use to understand traffic and behavior.
If you’re interested in implementing geolocation for your store or shopping mall, contact the networking experts at Eye-In Technologies
But to make the data truly trustworthy across the entire floor plan—especially at the perimeter—infrastructure design matters. Dead zones are not inevitable; they are often a sign that the AP layout was built for Wi-Fi coverage first, not for location geometry.
In Part 2 of this series, we’ll show how to fix that. We’ll walk through a simple, visual, step-by-step method to design an optimized AP mesh for geolocation—reducing dead zones, improving stability near edges, and making the resulting analytics far more reliable.