Creating an Engine for Predictive Retailing (aka “Nostradamus”)

A DHD Boot Camp Project

Developer's Help DeskAs a result of the 2019-2020 DHD Boot Camp, we built a new calculator that is intended to evaluate a membership’s relationship and status with the credit union, and based on a CU-defined set of parameters, present a list of possible things to sell to that member based on their unique situation.  We are dubbing this new engine “Nostradamus,” after the 16th century French astrologer and physician known for his predictions.

DHD Boot Camp Champions

Specifications for the first phase of this project were defined after numerous contacts with some very dedicated and hard-working participants from the 2019-2020 DHD Boot Camp:

  • David Jezewski, President/CEO, and Erin Beese, Marketing and Communications Director, Community Star CU (a CU*Answers client)
  • Karen Padrevita, Marketing Manager, Metrum Community CU (a CU*NorthWest client)

Overview

The project includes two major components:

  1. The Config:  A new configuration that includes a set of evaluation criteria that can be defined by the credit union, and a corresponding list of products and services to sell based on the results of that criteria.
  2. The Engine: A new tool that lets a user enter a single account base, run that account against the settings in the new configurations, and view a list of recommended products and services to sell to that member.

 The overall philosophy used during the design process was, “Do a psych profile and I’ll figure out what these people need,” as opposed to, “Find out what the person doesn’t have, make sure they’re eligible for that, then sell it to ‘em.”

For CU*Answers, the point with this DHD project is to evaluate whether the engine we build can actually produce a result that the credit union agrees with based on their own intuitive evaluation of that same member.  Before any predictive process can truly be integrated into any retailing process (a la “next suggested product”) via CU*BASE or online/mobile banking tools, we must first agree that a machine engine can evaluate a member in lieu of an actual credit union employee and come up with comparable results.

In this phase, the list of products/services to sell is a simple free-form text list.  This list does not tie to any actual CU*BASE product configs or tools, nor does it hook into any existing Cross Sales or Next Suggested Product features.  Depending on the success of the predictive engine, future phases will address hooking the engine to other features throughout CU*BASE and It’s Me 247.

Ideas for Future Phases

  • Add a “must hit” list for each profile, where the CU can define which attribute(s) must get a “hit” in order for that profile to be applied.  For example, for a “newbie” profile, the CU could define that the member must “hit” on the length of membership attribute (such as less than 60 days since membership opening).  If that criteria isn’t met, then the remainder of the attributes are irrelevant and the member would receive a score of 0 for that entire profile.
  • Provide a way to configure an alternate, member-friendly profile name to be displayed on scoring screens as well for future use via member-facing channels.
  • When running the calculation on a membership, show matched profiles only if they score xx% or higher, and if none are at that level, then display a generic “no suggestions” message in place of a list of suggested products.
  • Add an option on the calc engine to view suggested products for all matched profiles, not just the top scored one.
  • Add access to the calculator/results tool from various points throughout CU*BASE.
  • Create a report or dashboard that aggregates results for multiple/all members.
  • Build a more sophisticated engine for determining products/services to sell, including filters for qualifying members for products/services based on their current status (i.e., if someone isn’t eligible for a loan, don’t sell that even if their profile normally suggests that).
  • Hook up the engine and sales process to additional channels (It’s Me 247 or others).

Status as of June 2022: Nostradamus tools were released for active beta (field) testing with the 21.10 release, and participation in the test by all credit union marketing teams is highly encouraged!  In January the Asterisk Intelligence team held a 2-day meeting with the original project champions, to kick off the process of creating and testing the new configurations. Field testing is actively being conducted by both credit unions. A proposal for the next round of enhancements has been drafted and is being considered for inclusion in the 2023 development year.

Your chef for this recipe:  Asterisk Intelligence

 

Updated
June 3, 2022

Leave a Comment

* denotes required fields
  • (will not be published)

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

Check Out the New Recipes We’re Cooking in the Kitchen!

Check Out the New Recipes We’re Cooking in the Kitchen!

Have you visited the Kitchen lately?  If not, take a look at our newest projects: Card Activity Optics Credit Card Statement Enhancements Introducing Biz Watch for ACH: ACH Controls for Business Memberships Max Earnings Sweeps for Business Members Positive Pay Cashier Services (for Inhouse Checks) RDC Enrollment via CU*BASE Summary Statements for Business Credit Cards… Read more »

Sep 29, 2020

Check Out the New Recipes We’re Cooking in the Kitchen!

Check Out the New Recipes We’re Cooking in the Kitchen!

Have you visited the Kitchen lately?  If not, take a look at our newest projects: Accounts Payable Enhancements Creating an Engine for Predictive Retailing (aka “Nostradamus”) Deposit Hold Enhancements Escrow Analysis Enhancements Expanding Screen Sizes for CU*BASE GOLD Mobile First Transaction Limits for Express Tellers Each of these recipes includes a place to provide comments,… Read more »

Jul 14, 2020