Machine Learning: High on Hype, Low on Substance
Machines Can’t Learn and Predict without the Data to Teach Them (Good thing Xactly is sitting on 13 years worth of it!)
Machine learning is starting to feel a lot like the latest celebrity gossip—high on hype and light on substance. You can barely read a technology article or release today without someone espousing the latest machine learning (ML) and artificial intelligence (AI) capabilities.
But the truth is that any company can add machine learning to its platform; there is nothing particularly new about the technology itself. In fact, the idea of machine learning dates back to 1959, when computer gaming and AI pioneer Arthur Lee Samuel coined the term. His first test bed was using ML algorithms so that computers could learn to predict the best moves in the game of checkers. What makes machine learning so revolutionary today is the vast amounts of data that is now available and can be analyzed to form predictions.
Advancements in cloud technologies and online systems have brought about the democratization of data processing, opening up a bevy of new opportunities for AI/ML learning and insight. For the first time in history, companies are benefitting from what I call a “data trifecta.” Innovations in data processing have empowered companies with:
- The ability to capture data of all kinds, structured and unstructured;
- The affordability and vast compute availability for performance and scale; and
- The readily available ML algorithm libraries to fuel data insights and predictions.
These three factors together create an undeniable opportunity for any brand to transform not only its offerings, but also how it serves customers. Amazon is a perfect example. Over the years, it has learned about its customers, starting with the kinds of books they liked to read and later moving to merchandise, movies, and more. Its platform has learned patterns about user behaviors, enabling it to predict when and what they might buy and provide recommendations for other products to consider. This is only possible because Amazon has the access and rights to a vast treasure trove of data. According to a McKinsey AI report, Netflix applied ML to their proprietary data to improve customer search results, avoiding what could have amounted to a potential $1B revenue loss annually due to canceled subscriptions.
The same is happening in the enterprise world. But unlike Amazon and Netflix, many companies still don’t have the data behind their systems to back their machine learning claims. While enterprise software systems have been around for ages, former on-premises models resulted in data that was locked away in silos. The cloud changed that paradigm, enabling data across sometimes thousands of companies and users to coexist. For machine learning algorithms, this kind of data is an educational panacea.
The sales performance management space, in particular, is one where the hype of machine learning has not yet delivered on the hope of providing truly meaningful, actionable intelligence to customers. Again, it all comes back first and foremost to the data, as well as access to vast compute power and readily available ML algorithms.
Xactly, however, was born in the cloud. Our sales performance AI platform is backed by more than 13 years of aggregated and anonymized sales pay and performance insights from hundreds of thousands of subscribers. No other data set of this magnitude exists in the SPM market today. And with great data comes great possibilities and predictions that can optimize sales performance. For the past two years, we have invested significant resources in putting our data to the test, applying different ML algorithms and features ensuring we can deliver the intelligence our customers need at scale.
This week, we are proud to unveil the latest advancement from these R&D efforts, the first of several new machine learning infused capabilities in our Xactly Insights offering. Applying existing supervised ML algorithms to Xactly’s vast big data set, our customers will now be able to predict the probability of sales rep attrition. The prediction is based on numerous factors such as historical sales performance, years in the company, a sudden drop off in their pipeline, finishing off a great year, etc.
When a potential red flag is identified, sales leaders receive an alert directly in their Insights dashboard. They can then drill down into that alert and the history of a particular rep to see what further steps can be taken to avoid turnover, such as identifying new growth and coaching opportunities, simplifying plans, etc.
Knowing not only the high price tag but also the opportunity cost of losing a rep (especially a top performer), this level of proactive insight can be a game-changer. It can not only help companies maintain the health of their sales organizations, but also allow them to meet top-line revenue objectives and anticipate future staffing and hiring needs.
In any intelligent application, information is the key to learning, predicting and providing truly actionable insights. Unfortunately, for many applications in the market today, they are big on hype and light on information. Built on an AI platform and backed by more than a decade of real-world compensation data, Xactly’s solutions continue to drive the future of sales performance management with predictive analytics.
The sales rep attrition algorithm is just the first of many to come.
Complete Guide to Sales Team Compensation
Putting together your sales team requires you to consider the specifics of each person’s tasks, how those responsibilities move the team toward the given goal – and how to properly compensate each member.