Map the information before choosing a rule
A proximity pilot succeeds when it proves the technology works reliably within your specific physical environment, not simply because a beacon managed to transmit a signal. Before evaluating campaign performance, you need to confirm the infrastructure performed as expected. If the hardware or signal behaviour is unstable, any subsequent engagement data is unreliable.
The foundational metrics for a pilot are operational. Detection rate measures the percentage of time an opted-in device physically present in a zone actually receives the broadcast. If a visitor stands next to an exhibit for thirty seconds and the app logs a trigger only six times out of ten passes, the detection rate is 60%. Signal consistency tracks the variance in Received Signal Strength Indicator (RSSI) readings at a fixed point over time; wild fluctuations suggest interference or poor placement that will undermine zone accuracy. Hardware uptime monitors whether beacons dropped off the network due to battery failure, firmware crashes, or physical tampering.
A common mistake is declaring a pilot a success because the management dashboard shows thousands of notifications sent, whilst ignoring that the detection rate in the target zone was below 40% or that battery drain meant half the devices died within a week. Validate the plumbing before judging the water quality.
Defining KPIs for Proximity Pilots
Key Performance Indicators for a pilot must be separated into two distinct categories: technical KPIs that prove the deployment, and campaign KPIs that prove the business value. Conflating the two leads to confused post-mortems.
Technical KPIs should be tied directly to the physical deployment. These include broadcast interval adherence (did the beacon transmit at the configured 100ms, or did environmental factors force it to throttle?), packet loss in high-density areas, and the measured battery consumption rate versus the manufacturer’s data sheet. These are binary pass/fail metrics for the IoT integrator or operations manager.
Campaign KPIs depend entirely on the objective defined before the pilot launched. If the goal is queue management, the primary KPI is not notification open rate, but the change in perceived wait time or the reduction in queue abandonment. If the goal is exhibit engagement in a museum, the KPI is the completion rate of an audio guide trail. Writing down the primary business objective before switching on a single beacon prevents the temptation to move the goalposts to whatever metric happens to look favourable once the data comes in.
Measuring Notification Engagement
Measuring engagement requires understanding the strict funnel between a beacon broadcasting and a user taking action. The stages are: Sent (the platform issued the trigger), Delivered (the operating system received it), Seen (it appeared on the lock screen or in the notification centre), and Acted upon (the user tapped it). Vendors often report "sent" figures, which drastically overstate actual reach.
Operating system constraints
Engagement measurement is heavily distorted by the differences between iOS and Android background behaviour. On iOS, background BLE scanning is aggressive, but presenting a notification requires the app to have recent foreground activity or the user to have granted specific background permissions that Apple routinely restricts. On Android, background behaviour varies significantly by manufacturer and OS version, with aggressive battery-saving modes on devices from certain brands silently dropping background BLE connections entirely. When measuring engagement, always segment your data by operating system. A 5% tap-through rate on iOS and 1% on Android might not indicate a preference difference, but rather a delivery failure on Android devices.
Frequency and decay
Engagement metrics degrade rapidly with repeated exposure. If a visitor receives a notification every time they walk past a particular display, the open rate on the third exposure will be a fraction of the first. Measuring engagement without also measuring notification frequency will give a falsely optimistic view of your content’s relevance. Track the decay curve: how many triggers does it take before engagement drops below a useful threshold?
Measuring Conversion from Proximity Triggers
A conversion in proximity marketing is the completion of a desired action that originated from a physical trigger. Defining what that action is depends on the venue. In retail, it might be a voucher redemption at the point of sale. In a museum, it might be starting and finishing a themed audio trail. At an event, it might be navigating to a specific secondary stage.
The primary challenge is the attribution gap. If a beacon sends a push notification offering 10% off footwear, and the customer subsequently buys shoes at the till, connecting those two events requires integration between the proximity platform and the point-of-sale system. Without that integration, you are relying on correlation rather than attribution, which is notoriously unreliable in busy retail environments.
Where full system integration is not feasible during a pilot, the practical workaround is to use proxy conversions. Embed a dynamic QR code or a unique short code within the notification payload itself. If the user presents that code at the till or enters it at an event kiosk, you have a hard, attributable conversion directly linked to the proximity trigger. This limits the scale of what you can measure but provides data you can actually trust.
Benchmarking Against Industry Norms
Treating published industry benchmarks as reliable targets for your proximity deployment is a mistake. Vendor case studies frequently cite impressive engagement or conversion figures, but they almost never publish the conditions required to achieve them: the exact device mix, the opt-in mechanism, the notification frequency caps, or the physical layout of the space.
Consider two seemingly identical retail pilots. Site A has a high proportion of Android devices with aggressive battery management, concrete internal walls causing heavy signal attenuation, and a 30-second dwell time in the target zone. Site B has a majority of recent-model iPhones, open-plan retail fit-outs, and a two-minute dwell time. The notification delivery and engagement rates between these two sites will differ radically, rendering a shared "industry average" meaningless for operational planning.
Rather than chasing external benchmarks, establish an internal baseline. Use the first two weeks of a live pilot to understand your specific delivery rates, your audience's tolerance for notification frequency, and your natural engagement curve. Use that internal baseline as the control group. Measure subsequent success as a relative improvement against your own week-one data—testing a different message format, a revised zone boundary, or a new trigger time—rather than comparing your figures to an unverified statistic from a different continent, a different OS ecosystem, or a fundamentally different physical space.
