Moving from Coverage Predictions…to Coverage Facts
“Those who predict, don’t have knowledge. Those who have knowledge, don’t predict.” Lao Tzu
Lao Tzu was a Chinese philosopher and poet (although some sources cite him more as a symbolic figure) who first originated somewhere between the 6th and 4th Century B.C. That’s somewhere in the region of 2,500 years ago. He understood the fallacy of predictions – we make them when we don’t have the knowledge to be 100% confident in our assessment of either a present or future situation. No matter what we’re basing our predictions on, even with years of experience doing it, there’s still an nth percentage of “unknown” which can render our predictions useless in seconds. Have you ever been let down by a TV weather forecast? These aren’t the only predictions that can end in disappointment…
For years operators have relied on coverage predictions for an incredible range of use cases across their business. They’re utilised extensively by optimisation teams, supplied to regulators, and are even presented to the public via online coverage checkers, a trend proven to confuse customers who arrive on them searching for answers about live network experience. In all of these instances, wherever these predictions fail to line up with the reality they’re being applied to, there’s a problem not too far away. So how do we go about making coverage more representative and less predictive?
In essence, a coverage prediction is a model which factors in antenna height, clutter, terrain and more to determine the likely spread of coverage over a section of an operator’s territory. While it’s an excellent exercise in mathematical modelling, it fails to incorporate the more subjective and real-world experience of an actual person essentially saying: “Yes, I have 4G in this area.” On its own, this could be considered a coverage opinion. However, with the device reporting 4G connectivity in an area which the prediction agrees with, then you’ve validated your coverage prediction with a measured coverage fact.
These two data sets working in tandem can unlock the idea of validated coverage for a range of operator use cases but what about the future? Will there ever be a time when we can move away from coverage predictions entirely? Possibly not anytime soon. What can be achieved in the short term however, is the continual investment into crowd-sourcing real experience data from subscribers with a keen focus on integration, correlation and live validation against predicted coverage data sets.
Interestingly, current weather prediction modelling systems are taking a similar approach. The Met Office, Britain’s national weather service, began an initiative in the last few years to install small weather stations in volunteers’ gardens across the UK. This allows them to take their existing predictive data sets and validate them against numerous, ‘real-world’ measurements from across Britain. We’re currently helping operators around the world do effectively the same thing…using subscriber phones instead of weather stations of course.
If you want to make your coverage predictions more like a representative, living picture of experience, and less like a weather forecast, consider some further reading below or request some more information on our coverage validation approach.
Further reading: Why Coverage Checkers Aren’t Helping Your Contact Centre KPIs