Predictive NPS: a think piece on the future of customer loyalty

At the recent X4 Qualtrics Summit, our team at Fifth Quadrant attended a compelling session by Bain & Company exploring the rise of predictive NPS (Net Promoter Score) and what it means for the future of customer experience (CX). Originally developed in 2003 by Bain partner Fred Reichheld, NPS has long been a cornerstone of measuring customer loyalty and advocacy. 

The session underscored a significant evolution in NPS, from a reactive, survey-driven approach to a more proactive model, focused on anticipating the needs of the silent customer. 

Predictive NPS: The Future State of NPS

Predictive NPS can assess customer experience without relying solely on individual survey responses. Using predictive models, it can automatically estimate sentiment for all customers, through leveraging existing NPS data to assign a score to customers who haven’t directly participated in a survey.

It works by analysing customer behaviours and purchasing history, service wait times, and click patterns of responding survey customers, to estimate predicted NPS scores for non-responding customers. 

• Interaction behaviour: For instance, customers who call support 6+ times in a month and have long hold times or multiple transfers typically give low NPS scores. If others show the same friction points, the model will likely classify them as detractors.
• Operational indicators: Data like escalation rates, resolution times, or whether an issue was solved on first contact add weight. Customers who receive consistently smooth resolutions are often promoters, while repetitive problems push scores lower.
• Customer profile and transaction patterns: Sudden drops in usage, failed payments, or declining engagement can signal dissatisfaction. The model compares a customer to “lookalikes” in the NPS training data to assign a score.
• Sentiment from communications: Natural Language Processing (NLP) tools assess emotion and language in emails, chat logs, or calls. Phrases like “still not working” or “very disappointed” correlate with low scores, while positive resolution language can move someone from detractor to passive or promoter.

By identifying detractors or passives who wouldn’t typically respond to NPS, companies can proactively step in to improve customer experience by reaching out early, offering meaningful fixes, spotting bigger problems, testing what works. Similarly, identifying hidden promoters allows them to build relationships with these customers, aiding organic growth.

Looking Ahead

As customer expectations grow more complex, so too must the tools we use to understand them. Predictive NPS doesn’t replace traditional survey collection, it builds on it. Survey data remains essential as the foundation for training and calibrating predictive models. What’s changing is the ability to extend that insight across the full customer base, even where direct feedback is missing.

By combining traditional NPS with behavioural and operational analytics, organisations can shift from reactive measurement to proactive management, spotting risks earlier, identifying hidden promoters, and delivering a more responsive, data-driven customer experience.

At Fifth Quadrant, we’re big thinkers who thrive on exploring new ideas. We’re constantly in conversation about what’s next, challenging assumptions, testing new concepts, and reimagining what research can unlock for organisations. If you’ve got an idea, a question, or just want to explore some fresh thinking, we’d love to talk. Reach out, let’s start the conversation.