There is tremendous buzz about artificial intelligence (AI) and its power to transform business. Sales is no exception. AI is associated mostly with online selling at companies such as Amazon and Netflix. Lately articles have reported sky-high ROI from AI designed to boost field and inside sales force performance. From all the success stories, one might conclude that most sales forces are well on their way to realizing value from AI, and that the going is easy.
The truth is that only a small number of sales forces are using AI successfully. Some sales forces are stuck in multi-year efforts to build systems that have yet to see the light of day. And most have yet to take even the first step.
Consider one key application of AI in sales: increasing salespeople’s productivity in finding, acquiring and monetizing customers. (There are many other applications, but we will focus on this one.) AI promises to help salespeople generate and focus on higher-quality leads, gain insight into customer needs and preferences, coordinate selling with other channels (such as online and partner organizations), find opportunities to cross-sell and upsell, prevent customer defections and more. The reported results are impressive: higher sales, lower costs, big productivity gains, and a better sales environment.
The potential of AI to transform selling is huge. And the benefits are real. But the success stories you read about rarely offer enough insight about the challenges companies must overcome to realize the benefits. Consider four major challenges and some surprisingly simple but often missed solutions.
Bringing AI models to life requires new roles with specific skill sets. Team members from sales are critical for giving voice to the sales force, product and customer needs. The team needs big data experts, data scientists, machine learning engineers and more. To stitch all this together, the team also needs a boundary spanner—a person who understands both and sales and technology.
Such cross-functional AI teams are generally most effective when they come out of grassroots efforts, not from top-down directives. For example, when a technology company developed an AI system for one of its sales teams, the project champion came from within sales. The individual worked to cultivate a common belief and purpose for AI-enabled selling, bringing together more than 50 people across five business units.
AI systems work with vast amounts of data. Some of the data are structured (e.g., demographics, purchase history) and some are unstructured (e.g., words from emails or audio recordings). Assembling the data for a one-time use is difficult enough. Creating the processes needed to continually refresh the data can be daunting, time-consuming and expensive.
Fortunately, AI can work with incomplete or imperfect data, provided the data are free of systematic bias. In fact, AI can improve the quality of the data, for example, by predicting missing values or identifying possible errors. Do not let perfect be the enemy of good.
A casino wanted to use AI to optimize customer outreach. It had separate databases for tracking its high-value customers’ wagers, hotel spending, food and beverage spending, and entertainment spending. An estimate projected it would require six months of effort to integrate all these data, and even then, the data would be only 95% complete. So the casino chose to focus on a subset of data. In a couple of weeks, the casino had a 55% complete integrated database that it could update every month. By starting with partial data, the casino gained many insights quickly.
It’s intuitive that a complex AI implementation should involve agile, phased, and iterative design and testing, with ongoing feedback from the sales force. Yet repeatedly, we see organizations attempting to build comprehensive systems that are not sufficiently vetted by the salespeople and customers who can benefit. The error is often deadly. The resulting system is viewed as burdensome by a skeptical sales organization looking to see demonstrable value before investing to change the way it sells. Implementing AI should involve a series of sprints, not a marathon.
A biotech company attributed its success with AI to its decision to start small. The company’s broader vision called for an application that would suggest sales actions, content and messages for reaching customers across multiple sales channels. The launch began with a single sales channel. The rollout built on a progression of early successes to prove the approach was valuable before expanding the program more broadly.
Understanding and acceptance of any new sales tool or technology varies across salespeople. Many salespeople are naturally skeptical about the bold promise of AI.
Several strategies encourage adoption of AI tools by salespeople. To start, make sure the tools align with how salespeople execute their daily work, then deliver insights in the natural workflow. Provide explanations for AI-driven recommendations. For example, “This account is likely to buy this product because other accounts with similar purchase patterns eventually bought it.” Encourage salespeople to provide feedback about recommendations. For example, “Was the recommendation useful? If not, why not? Had you already addressed the issue? Or was it not a good idea?” Such feedback helps improve the AI system, but more importantly, makes salespeople partners in building the tools.
AI can have a significant impact on sales; the potential benefits you read about are possible. But don’t underestimate the time and effort required. By keeping things simple to start, remaining agile as you implement, and engaging the right team, you will see results.