Getting Started With Machine Learning for Sales Content

Human insight and inspiration are the basis of solidly profitable applications of machine learning to sales processes. LifeVantage, which recently launched a Gig Economy Group-based sales app for its distributors, improved its sales and marketing alignment as part of their onboarding to machine learning.

The GEG process builds on five principles that ensure clients onboard to improved, personalized sales experiences:

  1. Attribution analysis provides useful metrics for assessing content and messaging.
  2. Design to deliver “What’s Next” automatically, offering distributors the choice to use or refuse the machine suggestion.
  3. Make the process repeatable and scalable while supporting testable content and messaging variations.
  4. Deliver real-time leading indicators to management to facilitate their own content production decisions instead of blindly following machine recommendations.
  5. Provide a proactive view of the business from Day One.

We began the LifeVantage process using the company’s existing video, audio, and PDF-based content assets. Those assets served as the foundation of a machine learning platform. Assigning expected outcomes for each of these assets, whether LifeVantage management believed the asset moves a prospect toward purchasing, provided the GEG analytics system to identify opportunities for optimization. Each surprising outcome lets the machine learner choose alternative steps and test their efficacy in the funnel.

The teams assessed content provided to distributors as they join the organization, and consumer marketing content. Every aspect of the interactions between LifeVantage sellers and customers were examined and cataloged. With this inventory and a list of expected outcomes at each step, a machine learner can look at every interaction to identify what content works best, which messaging spurs prospect action, and the optimal order for delivering marketing content to customers.

Having innovated early with several technology vendors to develop four mobile apps, LifeVantage found its new distributor onboarding had become fragmented across several applications. Each app addressed different aspects of learning the LifeVantage Way, the selling process, and ongoing training. The company realized its distributors needed a single point of contact with LifeVantage information and its backend sales management systems. Simply providing training through streaming videos, audio programs, and coaching by sponsors would not be sufficient to keep a new distributor engaged if their early investment of time in LifeVantage did not convert to sales.

Results, Before The Machine Kicks In

The LifeVantage app for iPhone and Android connects distributors to the company’s media catalog and sales platform to provide next-step guidance for each phase of the selling process. Building on its established training and invaluable guidance from its most successful distributors, LifeVantage analyzed each step in its selling motion to create a machine learner capable of recognizing the salesperson’s progress in training, achieving product competence, and how each relationship they are working is progressing against network-wide benchmarks.

It was clear from the start a mobile-first strategy was essential. In many cases, distributors used paper-based sales management tools rather than PC applications. iPhone and Android versions of the app became priorities for the team, who moved to the design stage with a challenge: How to make the seller’s success as simple and pleasing as an Uber rider or driver’s experience.

A significant finding in early surveying and interviews centered on the design on framing seller’s choices. Distributors did not want to be told they have only one option, which they must follow the instructions given precisely. Instead, they sometimes want to skip sharing a video or time the gathering of feedback differently than established LifeVantage practice. For newcomers to LifeVantage sales, spontaneity in communication reinforced their sense of confidence. The team had to allow users plenty of flexibility. Experienced distributors moving from other network marketing companies were less inclined to watch training, but eager to get to work.

Through internal discussion and continuing interviews with experienced distributors, the team settled on critical metrics to change with the app that focused on novice sellers. While the first version of the LifeVantage app does offer services for long-term distributors, the newest seller is the next source of growth at the company. The selling steps and events that lead to initial success, such as time to first contact added and messaged, the frequency and success of the new distributor’s meetings, and speedy progress toward the first sale and progress up the compensation ladder became central to the project.

Customization At the Design Stage

Rapid development demanded continuous collaboration between the design and development teams during the onboarding and launch process. The initial distributor interviews produced a design the team decided required too much tool knowledge on the user’s part. The needed to know steps when adding a contact or creating a meeting, for instance, gets in the way of completing the task. A key decision resulted: Each action card that called for a distributor task should open the workspace where the work takes will be performed and, on completion, acknowledge the progress made. Sellers want feedback from their app about their progress.

App performance, too, presented challenges. Machine learning is an evolving computationally intensive technology. Building in a robust development environment on proven cloud systems was essential to fast responses to user input in the LifeVantage app. Feedback from distributors came fast and frequently, providing many signals about where to prioritize early investments.

The internal knowledge that had seemed concrete turned out to be merely intuitive guesswork in some cases. App usage demonstrated that new distributors wanted to get to work by starting to communicate instead of going through extensive training. Consequently, training sequences were shortened and selling steps moved to the “top of the deck” of the new user’s experience. Early usage showed distributors gravitate to creating new business. Consequently, LifeVantage recast much of its extensive media catalog as daily training material presented contextually while the distributor is performing a related task. When related to an immediate sales challenge, such as gaining commitments, LifeVantage’s training content proved even more effective.

Designers and developers worked closely to move the beta through seven release cycles, each shared with the beta community. Additionally, 20 top LifeVantage distributors and a cadre of 20 newly registered distributors were engaged to give weekly feedback. The result was hundreds of changes to the functionality and design captured over three months that would have taken a year to collect through traditional channels.

The first general release of the LifeVantage app delivered a simple-to-use tool for learning, selling, and supporting customer and distributor network relationships. It integrates the LifeVantage’s Media Library, Contact Management, Meeting and Feedback Management, as well as providing business performance information that assists sponsors in training their distributor networks. From first contact and entry of personal data into the app, through media sharing, it prompts distributors to make calls and send messages that capture prospect feedback. Based on what the prospect does, the app selects from a variety of optional next steps, such as triggering a distributor enrollment or shared cart with a potential buyer.