Better enterprise optimization algorithms
Enriching enterprise models doesn’t just improve price optimization. Many functions across the supply chain and virtually every business sector benefit by joining the identity graphs and enterprise graphs with machine learning models.
“One of the major problems we’re trying to solve here is using dynamic identity data at scale to drive smarter and more efficient operating metrics and more importantly real-time optimization models around your enterprise,” Velez said.
What distinguishes the platform from other solutions in determining more efficient enterprise metrics and models is its application of the customer identity. And most importantly for the data-science world, the platform is already connected with industry-standard identity graphs. Since over 60 percent of data-science effort is focused on data preparation, the Identity Applied Platform generates value much sooner. Publicis Sapient’s deep expertise building enterprise systems enables the reference architectures to connect customer identity and enterprise knowledge graphs.
Mitchell Weiss, the vice president of machine learning and data science solutions at Publicis Sapient in Boston, led the data science team that built the Identity Applied technology behind the platform.
“It’s about scaling a holistic intelligence solution where we’ve been successful with point solutions or installs or individual levels of embedding intelligence at an agency,” Weiss said.
He said it’s possible to connect the abundant signals being collected but that there isn’t a platform capable of arranging them in a consumable way. Every business has different technologies and different structures so there’s no silver bullet for aligning these composite elements resourcefully.
The Identity Applied Platform solves for this problem by approaching it from a different perspective.
“It’s about building the technology and platform that allows analytical and technical professionals to transfer and communicate intelligent solutions across platforms no matter what their origin is: what market, region or industry they are in,” Weiss said. “It’s the ability to make the solution itself the first-party citizen, not the technology that’s hosting or deploying it.”
For Weiss, it doesn’t matter if personalization is delivered through Adobe, Salesforce or some home-built solution. That’s not what’s important.
“The important part is knowing that we democratize the intelligent recommendation system that’s behind it,” he said.
How it came together
A confluence of premiere technologies and services formed the Identity Applied Platform. Built over the last five years, the cloud-native platform finds patterns within industries and supports dynamic, real-time models.
The Identity Applied Platform’s foundational technology is Epsilon PeopleCloud, an end-to-end measurement and activation platform that helps clients connect with people on an individual level. Epsilon PeopleCloud has unique data that enables direct access to predictive-intent signals that are not available anywhere else.
In 2019, Publicis Groupe, which helped build PeopleCloud, acquired Epsilon to expand its people-based marketing capabilities. Publicis Groupe used its deep technological and marketing resources to scale Epsilon PeopleCloud to find patterns within industries and generate predictive, actionable models.
Publicis Sapient CEO Nigel Vaz has been leading the Groupe’s strategy of how to bridge advertising and enterprise needs with transformation. In particular recognizing HOW the Identity Applied Platform’s modeling outputs could improve how companies across sectors conduct business internally: hotels could optimize room-block assignments, retailers could prevent customer returns, automakers could identify ideal pricing and incentive offers, customer and prospect identity drives smarter search and conversational experiences, etc.
“This is where we see the opportunity to differentiate and what’s missing for our clients,” Weiss said. “We are bringing a deep, rich understanding of customer identity, behaviors and data into those models as well.”