How accurate and falsifiable are your AI predictions?

There’s an incredible amount of excitement around us when it comes to AI. Consumer technology companies such as Google, Amazon, Uber, Netflix and Apple are using at AI at scale across every aspect of their business. Outside of a few examples B2B success adoption of AI by actual end users has been limited. We believe a huge part of the problem is that lots of companies with little to no AI under the hood have muddied the waters for buyers and users. At Aviso we’ve been building, selling and implementing our AI for Sales platform for over five years. In those five years we’ve learned a lot about AI technology, and we’ve also learned lessons in how to implement and get end users to actively use AI technology. We've learnt two important things working with our Fortune 500 and high-growth customers.

Number one: Accuracy is table stakes for users to embrace AI generated predictions 

End users and buyers want to see proof and demonstrable benefits before they invest time and effort in a new tool. With AI, we ask our end users to trust the predictions and insights that come from the AI with hard proof of their accuracy. Although there’s also been an incredible amount of talk around enterprise AI , there has been precious little talk about accuracy. At Aviso, we publish the results of our predictions for every quarter with our customers. In Q3 this year our WinScores had an average accuracy rate of 92% and our aggregate AI Day 1 forecasts were 91% accurate across our entire customer base.

Number two follows: Predictions should be falsifiable and companies should publish how they fared

Aviso bases all predictions on hypothesis that are falsifiable in principle and only makes claims when testing 100% false possibilities as not existent. None of the hundreds of players in Sales AI want to hold themselves to such a high bar because well, frankly, if a prediction didn't work out and no one heard about it, its ok right? WRONG. Predictions should be falsifiable. At Aviso we continue to back test and provide our customers with full visibility into the accuracy of our predictions. Read our press release here.