Predictive Analytics

Predictive analytics has been a cornerstone of Contemporary Analysis (CAN) since we first started over a decade ago. Prediction gives organizations the ability to “hit” more often than “miss”.

When you look at a predictive model it’s important to understand the outcome and how you measure success. It’s easy to say it needs to be 100% accurate all the time, but is that a valid metric for what you’re predicting.

At CAN we have built models that are extremely accurate and others that are “good enough” to see the appropriate outcome. The use case, and budget, will dictate the level of accuracy needed.

Examples of each:

Highly Accurate

HIPPA Compliant environment where an error results in a hefty fine. CAN built an attribution model for a health insurance company that has had only 1 error since it was built 7+ years ago. And even then, that error, proven with data, was shown to be a human error in selecting the appropriate outcome.

Good Enough

Prediction in a sales cycle. By narrowing the field of “likely to buy” prospects to the highest possible probability you can get a 60% conversion rate. So if your sales team calls 10 people 6 should say yes. To increase the odds they just need to increase the number of calls. Sales will close more often than they will lose a sale. Over time they will naturally find the patterns to become more successful and increase the closing ratio without additional predictive analytics.

Want to know which model you need? Setup up an appointment for us to learn more about your business.