Predictive models: How Accenture now predicts the probability of winning sales
October 7, 2019
Artificial intelligence (AI) is evolving from a hot new trend to an enabler of business transformation delivering real benefits within key enterprise business processes. As a technology-driven company applying “the New” now, Accenture is innovating by applying AI technologies and embedding insight into business applications to enable our people to make better business decisions. It’s what we call Applied Intelligence.
To help make this happen, we have a CIO Applied Intelligence team within our internal IT organization. This global team consists of business analysts, data scientists, user experience experts, and software engineers, and change management practitioners. We work with Accenture business stakeholders to shape, deliver, and operate analytics solutions that improve Accenture business process outcomes.
One area we’re focused on is predictive models. Predictive analytics, powered by AI, process Accenture’s huge volumes of data to suggest the probability of a business outcome based on a specific transaction and its similarity to other to the outcomes of previous, similar transactions. The model then presents those suggestions back to the end user in real-time, providing additional information to make better business decisions.
When we began our predictive model journey, our Accenture Sales & Pricing Excellence organization reached out to us with a potential use case. They wanted to improve sales effectiveness and drive better profitable growth. Specifically, they needed to help sales teams better understand if deals they pursue are really likely to convert into wins, or if not.
We suggested an AI solution that would predict the winnability of any given deal, at any point in time in the sales cycle, based on other similar previous deals that had either been won or lost. Together, the teams stepped up to this challenge.
Building the AI model
Our CIO Applied Intelligence team partnered with the internal IT team running our customer relationship management (CRM) system, Manage mySales. Using an existing, third-party artificial intelligence tool, we built a Predictive Model algorithm that we ultimately named Win Probability Predictor.
From there, an Accenture Sales and Pricing Excellence team joined the effort, and the entire team collaborated to train and improve the results generated by the Win Probability Predictor model. Drawing on five years of sales history, the Win Probability Predictor exposes positive and negative drivers of predicted win probabilities. The team enhanced the solution to give sales teams transparency of real-time probability scoring on deal opportunities as they work in Manage mySales. As teams adjusted deal characteristics, the model provided win probability updates. All of this helps teams see more quickly which deals to continue to pursue and which to stop.
What this means: An AI model can provide automatic and precise scoring of potential sales opportunities, in real-time, and on a global scale. It can provide explainable AI by showing the key drivers attributing to a winnability score. It can present this information to sales teams in a way that is easy to understand. Finally, it can embed the AI directly into the Manage mySales CRM system supporting the sales process so that sales teams can use this information to decide how best to proceed.
What It Does Today
Today, at any one time, approximately 45,000 sales opportunities are active in the Manage mySales CRM system, and every sales opportunity companywide is now scored by the Win Probability Predictor. The predictive model predicts the ability to win an opportunity with 97 percent accuracy—in less than three seconds. We continue to train this AI model on more than 120,000 new sales opportunities a year, so that win probability predictions continue to improve over time. Sales teams focus time and resources on client sales opportunities with highest likelihood of success. Accenture benefits from better profitable sales and revenue growth.
Advanced AI capabilities, like predictive models, hold the potential to be applied in any use case where scoring is beneficial. Other possibilities include revenue forecasting, risk assessment, sales campaigns, personnel scheduling demand and recruitment candidate matching. CIO Applied Intelligence continues work with Accenture stakeholders on new ways in which we can harness predictive model technology to powerfully impact our business. It is one core segment of our Applied Intelligence journey.
Accenture's Eli Lambert shares the different facets of how Accenture collaborates and innovates with SAP. Read more.
Sarah Dugan talks about the three-staged approach Accenture used to successfully roll out Microsoft Teams to its more than 500,000 employees. Read more.
Accenture’s Shane Marshall discusses how we are reimagining business services to better serve clients and enable our growth and resilience. Read more.