T1A

Intelligent Automation: Applying Next-Gen AI for Marketers at Scale

In Brief

It is becoming increasingly hard to step up the customer experience game in a market where consumers are digital-first, their journeys more diverse than ever and their decisions are driven by micro-moments rather than orchestrated outbound marketing action.
Coming up with the right sequence of actions and offers at the right time and the right place is now a make-or-break matter for the enterprise. Furthermore, it is not just about you and your customer, the pace of innovation and competition mandates immediate responses to changes and internal agility.
At T1A we identified the lack of solution that can fully address this challenge and provide marketers with a scalable self-learning solution. And thus we build an innovative product based on Reinforcement Learning which has already allowed one of our customers to achieve a 27 percent increase in campaign response rates.

Assessing What is Used Today

Imagine a major enterprise with a long-standing developed customer loyalty program. A program boasting a wide range of diverse campaigns generating customer interactions, 1–1 personalized offers and rewards. Sounds familiar?
While the major loyalty programs may be sophisticated enough to support an elaborate customer segmentation, most of them are still lacking the mechanics to really optimize the ROI in a true intelligent automated fashion. It is still manual cumbersome expert or statistically driven process which never results in a truly optimal solution across the board. Both approaches has obvious drawbacks.
Hence, a better product should meet all of the following:
● Easy to use, implement and maintain
● Best-in-class ROI given the business constraints
● Automatically adapt and self-learn in a rapid-changing environment
● Blending real-time digital data streams, 1st party offline data and existing expert knowledge

AI Journey of One Retailer

We faced this challenge while working with one of our major customers, a Global Top 50 retail chain from Eastern Europe called X5 Retail Group (think Kroger Co or Albertsons Companies, Inc.).
With a snowballing number of segments and offer types for millions of customers, choosing the most beneficial strategy to increase engagement eventually become a shady game of luck where you bet on analysts gut-feeling.
Our data science team, led by Valentin Marenich and principal data scientist Emily Volynets taken on the challenge to change that. They’ve went on a journey to develop brand-new next-best-action solution using truly innovative approach, that would be easy to use and at the same time be highly adaptable to specific business goals. On one hand, the system was meant to set a new straightforward process for surfacing and scaling insights. On the other, align with a company’s monetary KPIs. In addition, the decision-making process should be fully explainable to business users and C-suite.
We combined our expertise in retail and knowledge of machine learning techniques and developed a unique product that satisfies all mentioned criteria and shows an amazing profit. Reinforcement learning techniques under the hood enable self-learning and ease of deployment and maintenance. Furthermore, unlike most black-box AI products on the market, our system is transparent and interpretable, a perfect example of what is now being called Explainable AI.
“For us, digital interaction with customers is essential. We want to go beyond meeting their needs, exceeding expectations. Therefore, the predictive personalization of communication is one of the most important streams of the development of our brand, and the Reinforcement Learning tool provides us with valuable insights we need. We highly appreciate the results of the project and do not plan to stop there. In the near future, Pyaterochka’s [X5 Retail Group Brand] customers will find many more unique offers that will be based on a deep understanding of their needs and wishes” Galina Kardaeva, Head of Loyalty Programs.

What’s Next

Starting as a R&D project, we have managed to beat even the most optimistic ROI estimations, consequently becoming core CRM analytics tool, used by X5 marketers to automate testing and deployment of successful hypothesis, enabling them to finally focus on business and creativity.
Now, after months of improvements, usability and value testing, we are glad to announce that we would be offering this product to the market. An easy to use and deploy solution available for any business trying to bridge the gap between hoarding data and tangible business results through the power of next-gen AI.