How Traction is exploring critical questions to its business with artificial intelligence
Artificial intelligence (AI) is coming, and it’s often positioned in the media as either a massive wave of opportunity or an uncontrollable tsunami bearing down on humanity. How do you prepare for something like that? You’d be forgiven for reaching for the panic button, but in truth, being “AI ready” is a perfectly attainable goal — at least in a business context.
Discussions of AI tend to be hyperbolic and vague, making it difficult for most people to understand what AI is currently capable of and how far away we are from the angry wasp nest of utopian/dystopian buzz. To give our readers a better sense of what’s possible with AI in its current and near-future state, we thought we’d share some insights from our joint NSERC project with Simon Fraser University’s (SFU) Big Data program to explore AI use cases within our own business.
AI: What is it Good for?
There’s still a significant gap to bridge between today’s AI and autonomous vehicles (to name one popular example). For most businesses, AI’s greatest and most immediate value is in its ability to explore critical questions. To conduct our study with SFU, we needed to identify a number of questions we wanted AI to help us answer, including:
- How can we predict the probability of success in a given sales cycle?
- Can we forecast when a sale is going to close?
- Can we predict if and when a customer will attrit?
- Does positive employee sentiment lead to more successful projects?
AI has the potential to provide very powerful insights and follow through by putting those insights into practice. In Traction’s case, we not only want to know whether higher employee sentiment leads to more successful projects; we also want to assign employees to the projects that data tells us they’ll enjoy working on. That means happier employees, better projects and satisfied clients.
The Next Step: Preparing your Data
After identifying what we wanted to learn from AI, the initial stages of the Traction-SFU project have been to prepare Traction’s data for AI analysis. To, for instance, analyze whether or not a customer would interact with us within a specific time period, we didn’t just want to leverage data from our customers’ most recent activity. We want to leverage historical activity across time. We needed to build a model that would allow us to predict whether an opportunity would go quiet and when. A model that complex requires a good deal of preparation and planning. We’ve broken down the basic structure of how to get your data AI-ready into four simple rules:
- Ensure all data that’s supposed to be in Salesforce is in Salesforce
- Data from sources other than Salesforce should be clean and usable
- Carefully map all data sources and pathways
- Only process data that’s strictly necessary for analysis
“There’s a balance to find between collecting enough data to make insights valuable but not so much that you can’t process your data within a reasonable time frame,” says Mohamad Dolatshah, who lead the project’s first phase en route to completing his master’s degree. Mohamad has built the foundation of Traction’s AI initiatives by ensuring that data is well-prepared for all future analysis. “Being ready for AI is all about understanding your data,” he says. “If you aren’t collecting the right data to test the hypotheses you’ve identified, you won’t be able to make any meaningful inferences. Without that core understanding, machine learning loses its effectiveness.”
Salesforce and AI: Today and Tomorrow
In the cloud computing space, Salesforce and their customers are leading the charge to adopt AI. The Einstein platform (Salesforce’s AI product) is already available across clouds, delivering instant value out-of-the-box. Yet, with $1.2 billion invested in AI acquisitions and more than 175 data scientists devoted to increasing the sophistication of Einstein, Salesforce’s AI offering is only going to become more complex. Soon, AI will be applied to custom objects and apps, requiring adaptations to business process and consultation in order to deploy. Greater complexity requires greater preparation.
Are you Ready?
Your organization may not be investing in AI just yet, but now is the time to get your data AI-ready. By identifying the questions most vital to your business and implementing a data governance strategy that will keep your data clean and relevant, you’ll be better positioned for a seamless adoption of AI. If you need help getting ready, reach out to our team of data experts for a consultation.
“AI is a new science that finds answers in data,” says Mohamad. “Are you asking the right questions?”