Early detection

3 customer calls from the same area and you know you’ve got an incident so you’re out in the vans. Trouble is, that incident is already well underway by the time 3 different customers have decided that they need to call you. So here’s a different way of doing it: triangulation.

You call an incident based on 3 of the following occurring in the same or hydraulically related areas:

  • Event detection alert
  • Breached pressure threshold
  • Breached flow threshold
  • Social media
  • Customer call

A couple of those may benefit from a bit more explanation.

Event detection is the identification of unusual patterns in a time series dataset. It’s done by machine learning: computers build an understanding of what is normal from historical data, and then check all new data against that model to see if it fits the historical pattern or not.  When it doesn’t they throw up an alert. It’s something a computer can do efficiently, but you couldn’t do by eye even if you had the time and people to monitor every data stream all the time, which is of course an absurd proposition.

Customers are talking about you before they talk to you. Anyone else got problems with their water? Maybe a Tweet that isn’t addressed to you. Even if you could write a tool to spot social media comments, it’s not always obvious what people are saying because they can be sarcastic or ironic. ‘Great shower this morning!’ might not be a compliment. But there are tools that can interpret language and alert you to an issue.

It’s not a case of trying to decide which of these gives the best lead time and using that one.  Each situation is likely to be different. It’s about bringing all these together and increasing the certainty of there being a problem and doing so as early as possible.

And then you’ll be onto the incident much quicker than before, and be able to communicate proactively to customers before they call you.

i2O’s Network analytics solution is bringing together all of these data sources into a single place to give you the earliest possible lead time on an incident occurring. It will enable you to reduce leakage, customer minutes lost and customer complaints.