Geo-Fencing for Industrial Sites, Ports and Trade Shows

Roughly one South African mobile bid request in three carries a latitude and longitude precise enough to justify a two-hundred-metre fence. The rest is derived from IP address and resolves to a city centroid or a mobile network gateway, which in this market frequently means a device in Mpumalanga looks like it is in Johannesburg. Any agency quoting you fifty-metre geofences against local supply is either not looking at the bid stream or is counting the imprecise requests as hits. I build fences at the resolution the data actually supports and tell you what that resolution is before the flight is booked. Accuracy before ambition: The first deliverable on every one of these engagements is a supply survey: a sample of bid requests from the exchanges you buy, sorted by the precision of the location signal they carry, by source, and by device type. That survey decides the fence sizes. Where signal quality is poor, a polygon over an industrial precinct with a supporting first-party or contextual layer beats a tight fence that silently serves half its impressions to the wrong province. Fences I actually build: Polygons over plant and site boundaries rather than radii around a pin, because a refinery is not a circle and a radius drawn around Secunda or Sasolburg picks up a town's worth of residential traffic that has nothing to do with the operation. Port and logistics fences at Richards Bay and Saldanha where the audience is a specific set of contractors. Mining group head offices and the engineering consultancies clustered around them. Solar and wind corridors in the Northern and Eastern Cape where the addressable audience is contractors on rotation rather than residents. Event flights: Trade shows are where this technique is at its strongest, because the entire audience is inside a known boundary on known dates. Fences over the exhibition halls at Nasrec for Electra Mining Africa, over the CTICC during Mining Indaba and Africa Energy Indaba, drawn on the halls rather than the precinct so the surrounding hotels and traffic do not absorb the budget. Flights run from two days before to one day after, with a follow-up window against devices seen inside the fence where the platform supports it and your consent position allows it. Refused: Residential targeting of employees' home addresses, sold in this market as household or executive targeting. I do not build it, in South Africa or anywhere else, and POPIA is only part of the reason. Footfall attribution for a capital purchase. Store visit measurement was built for retail and means nothing when the decision takes a year and happens in a procurement committee. Fences under about two hundred metres against South African mobile supply, on the accuracy grounds above. If you want them anyway, I will show you the bid sample and let you decide, and I will not report on them as though they worked. Not for: Retail, restaurant and consumer footfall advertisers. Their measurement stack is different, their fences are smaller and the vendors who serve them well are not the ones I would recommend. Not for campaigns needing to launch inside a week either. The supply survey takes days and skipping it makes the rest of the work guesswork with a map attached. What you receive: The supply survey with precision by exchange, the fence set built as polygons and loaded into your platform, an event calendar with flight windows and boundaries drawn for each, and a post-flight read that separates impressions served against precise location signal from those served against inferred location, so you know what you actually bought.