Predictive Analysis Training: How Do You Earn Rep Trust?
Determining which action to take based on data can be difficult, especially for a salesperson who isn't actually diving into the data on their own. Using predictive analytics to help inform a sales team can prove to be an even bigger challenge, as you use data to see trends that will move the market needle in the future, but haven't happened yet. First, you need your sales team to trust the value of the data and the analysis processes. That takes careful communication and training.
Predictive Analysis and Your Sales Reps
Your sales reps probably don't want to be trained on how you actually do your job, or on the predictive model. But you can train sales reps to trust you and your data.
Though some salespeople are particularly data driven, says Blain Newton, executive vice president at HIMSS Analytics, those people are probably the exception. "For the ones that aren't naturally or automatically on board, earning trust in analytics takes time," says Newton. "It's really about proving the value."
A good predictive model — and a good presentation of that predictive model to your sales reps — isn't a "one and done" deal. Just because you got it right once doesn't mean you'll get it right every time. Data is forever updating. Consequently, you'll need to reassess your information week after week.
Let's say you're grading your sales team's pipeline, using insights gained from a predictive tool, based on likelihood that a prospect will buy. If you get that grade right, great! That's a point for you toward sales team trust in your methods. But if you get the grade wrong? If you say an account is a five-star account (with the highest likelihood to buy) and then it doesn't close, you'll need to follow up to understand why from a qualitative perspective, and to improve the analysis process. That will help you get it right next time and reaffirm your commitment to delivering real, actionable information to your teams.
"I spoke to a data scientist recently who said, 'all models are wrong. Some are just more helpful than others,'" says Newton. "So your model won't be right all the time, but you can go back and revalidate." You can speak with the sales rep on the account to understand why, exactly, a deal did or didn't close. Ask the rep why they might have won the deal and what drivers the decision maker expressed. You want to be sure that your predictions aren't just lucky, but instead, well-informed. Your sales reps' expertise in the field will help inform and improve the model.
But using these models well does demand some upfront trust from your sales reps. And to keep bringing in those collective wins, you'll need to start and continue a two-way conversation about why certain things work.
Look for Hidden Information in Your Model
Aside from asking your sales reps for feedback to encourage them, what else can you do to show the value of predictive analytics? Newton suggests pulling in the learnings from the model. Rather than just delivering a report on the pipeline grades, consider digging deeper to discover why a certain market is buying like crazy, or if price truly affected the outcome of the sale, for example. One HIMSS Analytics client discovered that win rates were much higher when there was a female decision maker. Uncover these nuggets of truth to deliver to your sales team.
"You should at least understand reasons behind success," says Newton. "Find out what your organization does well at and prioritize doing more of it." Your predictive analytics model can help you identify those successes.
Bring Your Reps Into the Decision Making Process
Another way to train your reps that predictive analytics should be trusted and are beneficial to the bottom line is to bring them into the discovery process.
"The best predictive models have domain knowledge and then color (specifics) baked in," says Newton. "Someone selling a certain tool has incredibly deep domain knowledge in that area. Bringing the reps in early to inform initial modeling on selling into a given area will be incredibly helpful." For example, the rep might know the details of the decision tree. Helping the rep define success means making the rep a part of that discovery effort. Your job is to let them see the results of building their thoughts into a mathematical machine.
What Do You Do When You Can Trust the Model?
We've talked a lot about how you can get your reps on board with predictive analytics. You can involve them in the discovery process, ask them for qualitative feedback, and provide valuable information about correlations you find as you work toward full trust.
"At some point, though, the model will be defined enough that you can absolutely trust it," says Newton. The trust, then, will need to extend to your recommendations. You'll know that your reps will never win a certain deal, and will have to eliminate those deals from the pipeline entirely. That can be a hard conversation, so you'll need to come equipped with the right information. You need to be able to say, "You have never won a deal of this type, and we can show you exactly why you're unlikely to win it."
Having these conversations once every six months won’t help you build trust. Look at your accounts every single week and see if there are changes. It’s not just based on the data you have at the beginning of the month. You may find that some action occurs in the third week of the month because people become active buyers at the moment they start researching tools, for example. So you need to be agile and aware enough to reassess constantly.
"That’s all you need. It needs to be iterative, and it needs to be rapid iteration," says Newton. To continue to prove to your reps that your model works and that you're engaged, continue to ask questions about what's happening during the iteration. If you see an account move, ask questions about why.