⚙️ Product updates

Predictive Maintenance

You have probably already heard during the October Sharing is Caring all the things that our team is doing to become the HEMA t-shirt of condition monitoring.

In case you missed it and want to know more, you can check the presentation by following this link.

Savings

Finding potential savings for our customers is great! But knowing they took our advice and implemented them is even better. In the past month, we wrapped up a series of efforts to get a better feel for this implementation coverage. So far, we did not know if our customers did something with our advice. We don’t often hear back from them after the energy scan, and there are no clear trends if we look at overall consumption. We were afraid that the hard truth was that customers did not act upon our advice. During this year's spring, we thus decided to ask them: what did you do with our advice and why? We collected a lot of valuable insights for our consultants. Customers certainly did use our advice in practice, leading to sizable reductions in their energy waste (standby, outside production & protocol deviation).

This effort not only resulted in these good and comforting insights but also provided us with every data scientist’s dream: a labeled dataset. Having a subset of customers for which we certainly know if they acted upon our advice or not allowed us to apply some statistics.

We could test different ways of measuring impact and try different strategies for flagging implementations. This is valuable information to construct company KPIs and determine how we can motivate our customers to do more! We can notify them of detected energy waste reductions and tell them to keep up the good work, or we can remind them about the advice at the right moment. If we also allow them to respond to these messages, we get more labels to get a better estimate of implemented savings.

How we will do this is something our new Product Manager will look at. He will research what type of interaction motivates our customers most and how we can get the most valuable feedback. We still have to wait a couple of weeks before this Product Manager will start, but with the outcomes of this effort, we can provide him a head start!

Smart alerts

The data team has worked hard to improve how we detect standby power consumption of our customers’ machines. Now that these improvements are in place, we are confident that we can accurately identify most standby. We already use the standby algorithm during the Energy Scan and in our weekly Tracking Reports, but can we do more with it? Can we show standby waste in the web app or even use the algorithm in our Alerts service?

The main goal of the Smart Alerts project is to be able to detect standby in real time, so that we can make our alerts smarter and the web app more informative regarding energy waste. To do this, we need to rebuild part of our platform to be able to analyze streams of data better in real-time. An additional benefit is that the new technologies we use for this are faster and more scalable than the existing ones. This will significantly help us improve the customer experience and reduce the number of incidents as our customer base grows.

During Q4, our goal is to set up these new streaming technologies, process the raw energy data in the same way as we did in the old platform, and create the first MVP of real-time standby detection. We are currently on track with rebuilding the processors and ingesting the historical sensor data into the new platform. A key challenge is to limit the customer impact: making sure that the data still looks the same for the customer when we eventually make the switch.

There will be a demo of this exciting new project in the Sharing is Caring early Q1!