Hi Nik,
Looks like you’ve already got enough responses to write the book, but here’s my two cents on our experience if useful, mostly at museums and indoor cultural institutions , plus a couple of outdoor sites too. Detecting presence using WiFi relies only on the visitor having a device with WiFi turned on. They don’t need to connect to the network (though this helps data quality), don’t need to download an app, accept permissions, be actively looking at their WiFi list, have ‘ask to join’ on, etc. In our experience, WiFi enabled devices are far more pervasive than Bluetooth. Detection works because as our devices are surveying the field for possible networks, they send out a signal in doing so, their MAC – a bit like a game of Marco Polo. A MAC address is non-identifiable, ie we can only see this is device 1234, not Nik’s device, with no reverse look up facility. However, privacy is still important – data should be viewed only in aggregate and the raw address should not be access ed by general users. If required, the MAC can be hashed at source, though this reduces data quality. Detection can be performed by the venue’s own WiFi network (most mainstream vendor s e.g. Ruckus and Cisco as mentioned offer this facility), or with a hardware accessory. Ranges vary, and some can be adjusted to scan an isolated area. Battery power is possible, but not desirable as it does require more power than say a b eacon, and devices only last a week or so before having to be recharged which is logistically difficult. They can be hidden from sight, but perform better at height and also need to be weather proof, which can be tricky outdoors. These also still require a WiFi connection to report back remotely. We then connect via API to stream this data, combining it alongside other relevant data sources, such as online, social, transactions, weather etc. This raw data as you’ve probably discovered needs to be clea n ed up; before insight analysis and then visualization. In our case, we’re accounting for all sorts of influencing factors. These differ city to city, site to site, but a very rough guide is 92% of visitors carry a device to a cultural institution (may differ for parks than museums), 75% WiFi enabled. Then, you have to allow for everything else, people who carry multiple devices, randomization, fixed equipment, staff movements, etc. Outdoors, your configuration would need to protect against passers by and traffic. Devices differ in their advertisements, network vendors differ in their scanning reach and intervals, both impact accuracy. Outdoor environments also play a part, especially if there are a lot of trees . Parks are tricky as often there are no defined entrances and therefore the coverage has to be high unless deemed a sample only . A sample manual count can help validate the overall scaling factor. Where possible, we rely upon machine learning as part of this cleansing, important because other than differing by site, these factors rarely stay the same over time. Using this approach, we’ve managed to get accurate results when compared with other counting methods, and generally speaking each method has its own flaws (clicker counters case in point). Additionally, WiFi data can reveal zone activation, trail routes, dwell times, repeat visitation etc. I imagine this data would be equally valuable to the park. Some of our clients use presence to understand visitor behavior and take their visitation counts from ticketing or elsewhere, others rely upon presence to report visitation itself, as well as behavior. At parks, the question then quickly turns to ‘what constitutes a visit ?' , especially with thoroughfare. For example, if I cut through on my way to work, am I visiting? If I do the same on my way home, is that two? I would imagine the value of this over and above validating the big number, is to start to see the impact of levers within the park’s control, and influences to otherwise allow for, in both overall visitation and engagement. This insight can then be tracked against the capital plan or used for operational purposes. For example, promotion, digital engagement, seasonality, weather, events, what’s on in the area, new facilities, maintenance, etc. And lastly I might add, though we have yet to measure the angle of curiosity, rest assured we’re working on it. Angie Angie Judge Dexibit www.dexibit.com an...@dexibit.com @Dexibit #musedata
_______________________________________________ You are currently subscribed to mcn-l, the listserv of the Museum Computer Network (http://www.mcn.edu) To post to this list, send messages to: mcn-l@mcn.edu To unsubscribe or change mcn-l delivery options visit: http://mcn.edu/mailman/listinfo/mcn-l The MCN-L archives can be found at: http://www.mail-archive.com/mcn-l@mcn.edu/
