We Check the Customer's Pulse Before It Turns into a Complaint
The most insidious type of customer loss is the customer who leaves without making a sound. Dissatisfaction rarely starts loudly; it manifests itself in a gradually hardening tone in emails, WhatsApp, or live support, and turns silently into churn when unnoticed. Phexum Pulse is an NLP-based customer experience and churn prevention platform we developed precisely to make this silent risk visible. Used for years by the financial technology leader Multinet Up, Pulse works like a smart radar catching dissatisfaction before it turns into a complaint.
Featured Results
Generating value in the Multinet Up ecosystem for years, Pulse has achieved concrete KPI successes. The average response time to customers whose pulse rose—meaning those entering the risk group—dropped from 2 hours to 30 minutes. The rate of high-tension, harsh communications dropped by 3 times. An absolute 55% decrease was achieved in the overall customer complaint rate. These results are the measurable equivalent of intervening in customer dissatisfaction not too late, but before it grows.
Who the Client Is and the World They Live In
Multinet Up, founded in 1999, is Turkey's leading next-generation financial technology and services company; offering payment and loyalty solutions that add value to institutions, their employees, and consumers. For a fintech at this scale, customer communication means massive and continuously flowing data.
The fundamental reality of this world is this: swift and quality responses to customer requests directly increase satisfaction; when requests are carefully resolved, customer churn decreases significantly. However, the biggest risk is the customer who silently experiences dissatisfaction and is lost because it goes unnoticed in time. In written channels where customer communication flows most intensely—namely email, WhatsApp, and live support—the longer the solution is delayed, the harsher the language becomes, and the bond weakens. It is impossible to track the "tone" of every message in such a dense flow with the human eye; Pulse was born exactly to eliminate this operational blindness.
Cross-Cutting Concepts and Architecture
The first concept running through this entire platform is pulse, that is, translating a customer's emotional state into a continuously measured live signal. The first leg of this is understanding the written text. Its technical name is sentiment analysis; however, we broke the "positive/negative" binary mold the industry is used to. We built a Machine Learning Model that instantaneously parses every interaction coming from written channels using NLP and classifies the text's degree of negativity across 6 different depth levels. The benefit is this: instead of a two-ended label, you get a fine scale measuring exactly how tense a customer has become, which clarifies how urgently a case must be intervened.
The second concept is proactivity. An emotion scale is valuable only when turned into action. Therefore, we placed a prioritization and warning layer on top of the system: the most intense requests are automatically detected and moved to the top of the customer relations team's screen, and when the customer's pulse reaches a critical threshold, an instant warning and action report go to the relevant managers. The technical equivalent of this is churn prevention: instead of reporting customer loss after it happens, preventing it before it occurs. Pulse also provides root cause analysis by clustering the topics where complaints are concentrated and automatically reports metrics like NPS and CSAT with its internal survey module; thus, while managing instant crises, it simultaneously builds the ground for permanent improvement.
How the Collaboration Began
Multinet Up had a high-volume, continuously flowing customer communication; the need was for a system that could instantaneously see the risk within this flow without human intervention. As Phexum, we developed Pulse end-to-end. This process encompassed algorithm design, text analytics modeling, CRM and CX integrations, and product management. The goal was not a reporting tool, but to establish a radar working live inside the operation, bringing the risk instantly in front of the team.
Building Blocks of the Solution
Pulse works by connecting four layers. The NLP and 6-level sentiment analysis layer instantly parses every interaction from written channels and classifies it according to the degree of negativity. The smart prioritization and proactive warning layer moves the riskiest requests to the top of screens and generates an instant alarm for managers at critical thresholds. The CRM and CX integration layer works fully integrated with the company's existing systems, writing the generated insights back there; on a special customer profile screen, every customer's digital health status and instant pulse can be tracked over a timeline. The root cause and survey layer clusters complaints by topics and automatically reports critical metrics like NPS and CSAT.
Instead of reporting customer loss after it happens, preventing it before it occurs.
Awards and Sectoral Recognition
The AI-supported customer experience Pulse provided to Multinet Up has been awarded by national and international authorities:
- Brandverse Awards (2026): Bronze Award in the "AI-Supported Customer Experience" category.
- Istanbul Marketing Awards (2025): Gold Award in the "Digital and Technology / AI Usage" category.
- PSM Awards (2020): Gold Award in the "Innovative Customer Experience Interaction" category.
- Sardis Awards (2020): Silver Award in the Innovation section "AI Usage" category.
- AI Tomorrow Summit Awards: "CRM Customer Relationship Management Award" under "Projects Adding Value to the Future".
- CX Awards Turkey & BoSA Awards: Three grand prizes in these customer experience-focused competitions — "Most Innovative Special Jury Award", "Best Use of Technology in Customer Experience (Big Idea Award)", and "Customer Experience Category Award".
The Impact We Created
Pulse transformed into an infrastructure that enables Multinet Up to maintain its bond with its customers. Customer satisfaction, which is more important than anything for Multinet Up, can now be prioritized before reaching the complaint level, and permanent solutions can be produced to eliminate dissatisfaction. The shortening of response times, the reduction in harsh communication, and the drop in the complaint rate are the measurable sides of this transformation; the unmeasurable side is noticing and retaining the silently leaving customer in time.