Moving from Descriptive to Prescriptive Analytics
Prescriptive Analytics in Insurance
A simple Google search on the different phases of Analytics returns similar results – a graphic with Descriptive, Diagnostic, Predictive, and Prescriptive Analytics.
Here’s how each can be summarized in a phrase:
- Descriptive – What happened?
- Diagnostic – Why did it happen?
- Predictive – What will happen?
- Prescriptive – What should we do about it?
Let’s take the Analytics Maturity Model and tie it into a real-world example like buying insurance from your local State Farm or Nationwide Agency.
Only a few years ago your experience might go something like this – jump on the Internet, Google “cheapest auto & home insurance,” and schedule a time to meet with an agent.
You end up with John Smith, one of the least effective agents in that particular branch (sorry, John). During your meeting, you review your home and auto bundle options, run a few quotes, and ultimately decide to NOT move forward with a policy.
Even an average agent will recap the meeting after you leave to uncover what happened (you didn’t buy) and why it happened (could be one of a million different reasons)? The “why” is far more subjective. What the agent doesn’t know is that you were just price shopping looking for the lowest quote to beat your current policy with another carrier.
Replay the same scenario with one of the top-performing agents Mary Jones. Mary is peeling back the layers of your buying intent during your meeting and working further down the Analytics Maturity Model. She’s not asking “what happened” or “why did it happen” after the meeting. She’s asking herself “what will happen” and “what should we do about it” during it.
She’s reading your behavioral queues in real-time and adjusting the conversation to optimize the experience based on your specific needs. The best insurance agents or salespeople are Predictive and Prescriptive. They adjust “on the fly.”
Now let’s think about another “salesperson” most businesses have that is fairly static and treats every prospect and customer the same outside of some A/B testing….your website.
Just like agent John Smith, chances are you have tools at your disposal that are Descriptive and Diagnostic. They tell you what happened (heat maps and funnels) and then let you infer why it happened because, like an in-person interaction, it’s nebulous. You go to your development team, adjust your website, and rinse and repeat the same process over and over trying to squeeze out a basis point of extra conversion rate.
You’re probably asking yourself, I already have a chatbot and email campaigns to “prescribe” an experience and interject in the process if a user is abandoning. But is a pop-up after 7 seconds or 50% scroll, or an ever-present Chatbot really Prescriptive?
No, it’s not. It’s static and the same for everyone. It’s as if you forget every other user you’ve seen on your website each time a new one comes to it.
So what’s missing?
What’s missing is the third step in the Maturity Model: Predictive Analytics. Why treat every user the same when you don’t have to.
Create Dynamic Experiences by Combining Predictive and Prescriptive Insurance Analytics
Let’s think back to the insurance agent Mary Jones. Mary was reading behavioral cues and adjusting the experience individually for each prospective policyholder. She did not have a one-size-fits-all approach that was the same for each prospect.
Based on her experience with other prospects, she learned patterns of prospects who buy or don’t. As she sees more situations, she applies that knowledge to the next prospect thus getting better each time.
And while these experiences lead to greater expertise, Daniel Schreiber, Founder of Lemonade, said it well, “What we’re seeing here is something that is going to be very traumatic for the whole insurance space. Data is overtaking expertise.”
So how do you combine Mary’s expertise with big data and turn 2+2 into 5? With Behavioral Intelligence.
So much of what “artificial intelligence” is doing these days is replacing tedious, repetitive tasks done by humans. More advanced AI is actually learning and improving on those tasks so the output is not only faster but better than the job done by humans.
As AI advances and technologies like Behavioral Intelligence become mainstream, machines are becoming more and more capable of reading, analyzing, understanding, and reacting to situations that used to be inherently human.
The digital transformation opens the door to new possibilities in the analytics field because it is the great equalizer of experiences.
This makes it possible to compare uniform experiences across an exponentially greater audience, predict what their intent is, and then “prescribe” different paths for the individual user.
The traditional 20-150 data points collected when an application is submitted grows 100-fold when using Behavioral Intelligence. Every nuance of digital behavior is analyzed, providing a much more granular picture of the applicant and their accompanying risk. This makes accelerated or automated underwriting and straight-through-processing possible.
By combining big data and behavioral analytics, predicting risk and quantifying losses is a much simpler process and turns your website into your greatest salesperson
Just like Mary Jones, Behavioral Intelligence allows you to take a uniform, static experience and make it adaptive and dynamic. With real-time predictive analytics, applications come alive, and intelligently react to the user’s behavioral cues to encourage the best outcome.
For instance, if a user is trending towards abandoning the application, inject your “Live Agent ChatBot” at the optimal time or present them an offer to encourage completing the application. Or if a user is trending towards a “high-risk” score, dynamically add friction to ensure you are pricing the policy as effectively as possible.
The possibilities are endless and the data is there, you just need the right tool to utilize it to its fullest potential. That tool is ForMotiv.
Email us to see how we can help your organization move from descriptive to prescriptive analytics.