Improving customer encounters with AI: the PATA project

When I started working on the PATA project, I was excited by the opportunity to apply machine learning (ML) in the context of real-life communication between salespeople and customers. My role in the project focuses on developing ML models in close collaboration with participating companies and fellow researchers. What makes this work especially interesting is the combination of sales expertise, data analytics, and practical business needs, all grounded in real interaction data.

In today's fast-paced sales environment, artificial intelligence (AI) is often viewed as a tool for automation. But what if AI could become something else entirely, a smart, invisible partner that helps salespeople listen better, communicate more effectively, and connect with the customers at just the right moment?

This is the central question of the PATA project (AI-Enabled Customer Experience), led by Haaga-Helia University of Applied Sciences and the University of Turku and funded by Business Finland. The project explores how AI technologies, from ML to speech analytics, can enhance human interaction rather than replace it. The goal is to support smarter, more impactful customer encounters that still feel personal and relevant.

Speech Analysis in Sales Interactions

One area of the project that I've found particularly fascinating is the use of speech analysis to better understand conversation dynamics. This includes analysing pitch, pacing, rhythm, and turn-taking to identify patterns in effective and ineffective communication.

Using these features, AI systems can detect conversational behaviours such as interruptions, monotone delivery, or excessively long turns. In parallel, the same analysis can highlight positive features like varied intonation, balanced pacing, and engaged responses. These insights support the improvement of sales conversations and the development of essential communication skills. Furthermore, the data can be used to identify personal communication styles associated with better sales performance.

Developing Predictive Models for Sales Outcomes

Another major focus of the project is developing predictive models that help make sense of complex customer data. By analysing past interactions, behavioural patterns, and purchase histories, the models aim to estimate the likelihood of a successful sale.

These models are intended to help sales professionals allocate their time and efforts more effectively by identifying the most promising customer engagements. They also enable more personalised and timely communication strategies, replacing guesswork with data-driven decision-making to improve sales outcomes and strengthen the customer experience.

Dynamic Seller Profiles for Personalised Coaching

As part of the project, we are also planning the development of dynamic seller profiles based on real interaction data. The idea is to group salespeople according to communication styles and behavioural patterns, providing a basis for more tailored coaching and development.

The aim is to enable training and feedback that can be aligned with each salesperson's communication habits, increasing the relevance and impact of coaching efforts. From a research perspective, this approach demonstrates how ML could support more targeted and individualised development tools in sales contexts.

Cross-Industry Collaboration

The project is conducted in close collaboration with companies such as Rainmaker, Telia, Helen, Fastems, 3StepIT, and LeanLab. These partners contribute real-world data and insights, which are essential for testing and refining our models in practical environments.

Collaboration with these companies ensures that the research stays connected to operational realities and that the tools developed address actual business challenges.

Tanja Vähämäki
The author is a doctoral researcher at the Faculty of Technology, University of Turku and a project researcher in the Pata (AI-Enabled Customer Experience) project. She examines how companies can leverage their data and apply AI in ways that support business development and operational efficiency.