Submitted by Erik Udstuen, CEO & Co-Founder at TwinThread
With any predictive solution, results aren’t guaranteed. Predictive analytics applies a scientific methodology that suggests it is best practice to make decisions based on the best information at hand. However, as much as we’d like it to be, it’s not a crystal ball.
So, if there isn't any solution provider that can guarantee results, how are you supposed to figure out whether more profit can be squeezed from your operations? The key isn’t just in the usability of the platform (as important as it is). It’s also in the platform’s ability to be quickly implemented and applied within your unique environment. A slow proving process is not a scenario you want to subject your organization to.
Imagine putting large amounts of time, effort and investment into a proof of concept for the implementation of predictive analytics, only to find that the insights derived from the platform are of little or no worth to your domain experts. From a business perspective, it could be disastrous – and, from a cultural perspective, it will most certainly be discouraging.
The truth is, if presented with such a risky proposition, most businesses would have to make the decision to pass. How could they be blamed? However, let’s be clear, you’re not proving technology in a production pilot - you’re establishing time to value and the potential to scale.
You should be empowered by your predictive operations provider to find out fast whether their solution will make the grade for your operations. In a brief period, you should be able to make an educated decision on whether it’s time to get out or it’s time to double down on the use of the solution.
If it’s not going to work, you won’t waste your time expending any unnecessary effort – reducing residual cost. If it’s clear that there is value to be had through the use of the platform, because of the speed to value, you’ll be able to expedite insight to action – increasing profitability.
Your organization and its innovative members just want to improve – you shouldn’t be made to wait to see whether a particular solution can help you accomplish that. Ensure you constrain your pilot for fast success or fast failure. Always begin with the end in mind - how will model results be operationalized and at what speed?