How Pledg built a customer feedback loop that runs itself
Pledg closed its Customer Context Gap — turning scattered surveys, support conversations, and reviews into an always-on intelligence layer that reasons across all of it and delivers what each team needs, automatically.
The challenge
Pledg is a French fintech making Buy Now, Pay Later seamless for merchants and their customers. As demand grew, so did the volume of customer feedback — survey responses, support tickets, sales conversations, online reviews, and interviews.
Pledg's dashboards showed what was happening: conversion rates, ticket volumes, NPS trends. But none of that explained why. Why are merchants hesitant to adopt a new payment flow? Why are support tickets rising in one segment but not another? Why did a promising feature underperform?
The answers were buried in what customers were actually saying — across channels, in different formats, at high volume. The team tried manual synthesis: reading comments, tagging themes in spreadsheets, writing up summaries. But that approach couldn't keep up. Insights arrived late, reached the wrong people, and often amounted to lists of verbatims with no clear direction.
As a regulated European fintech, Pledg also needed all of this to happen without compromising on data privacy or GDPR compliance.
What Pledg needed wasn't another tool to look at feedback. They needed a system that could reason across all of it — continuously — and turn it into priorities each team could act on, delivered where they already work.
Results at a glance
From fragmented signals to one intelligence layer: Customer feedback from every channel is continuously processed into an interpreted, structured layer — not raw dumps, but reasoned outputs teams can act on.
Autonomous distribution: Insights are compiled and delivered to each team automatically in Slack. No one has to chase, compile, or wait for a reporting cycle.
Fewer mis-bets: Teams pressure-test priorities against real customer evidence — ranked themes backed by volume and quotes — before committing resources.
Token-efficient, instant answers: Because NEXT builds an interpreted intelligence layer from the raw data, every query is fast and reliable — no context-window overload, no running out of tokens mid-analysis.
"With NEXT, we strengthened our understanding of our customers, competitors, and market. It gives us deep actionable insights we use every day to improve our product, marketing, and overall strategy — while meeting our data privacy and security standards."
– Adrien Bonhomme, Chief Product Officer
Inside Pledg's Customer OS
Pledg uses NEXT AI as its Customer OS — the intelligence layer that bridges the gap between transactional data and real customer context:
Unify — Surveys, support data, reviews, and sales conversations flow into one system, organized around the people and accounts behind them.
Understand — AI agents continuously process incoming feedback: clustering themes, quantifying what's growing or shrinking, surfacing pain points, desires, and objections — not just keywords, but meaning.
Act — Intelligence is pushed directly into Slack channels and DMs, routed to the right team by topic. Insights also feed product backlogs, marketing priorities, and competitive positioning — for humans and AI agents alike.
Connected sources:
Survey responses
Support tickets and conversations (Zendesk)
Online reviews (Google Business)
Sales calls and customer interviews
Pledg's signature workflows
1) Always-on feedback distribution — insights find the right team
The job: Eliminate insight lag. Get what matters to the right people — without anyone compiling reports.
How NEXT does it: AI agents continuously process incoming surveys and support data, reason across the themes, quantify what's changing, and post team-ready updates into the right Slack channels and DMs. Each team gets what's relevant to them, when it matters — not a raw feed of verbatims, but structured intelligence with ranked drivers, evidence, and clear direction.
The result: Decisions happen faster. Coordination overhead drops. Signals don't get missed — because intelligence shows up where work happens, automatically.
2) Product + support loop — stop recurring issues at the source
The job: Turn repeated tickets and complaints into product fixes — not just support responses.
How NEXT does it: Teams ask questions like "What are the top recurring reasons customers contact support this month?" or "Which issues are rising fastest — and what changed after last month's release?" NEXT AI reasons across all the support and survey data — not a sample — and returns ranked drivers with counts and representative quotes. Product can see exactly what to fix and why, grounded in the full picture.
The result: Less rework, fewer repeated issues, and a tighter loop between what customers experience and what gets built.
3) Decision validation — reduce mis-bets before shipping
The job: Pressure-test roadmap bets before investing time and rollout effort.
How NEXT does it: Teams validate assumptions — "Will enterprise customers benefit from this feature, and why?" — and NEXT surfaces supporting signals, counter-signals, and segment differences. Because the intelligence layer is pre-built and interpreted, answers come back in minutes, not hours, without burning through token budgets.
The result: Better prioritization, fewer wrong bets, and faster iteration grounded in evidence across the full dataset.
Why NEXT AI — and why not just ChatGPT
Pledg's teams already had access to general-purpose AI. The gap wasn't the model — it was the data layer underneath.
Pasting feedback into ChatGPT means working with raw, unstructured data in a limited context window. You get confident-sounding answers built on a tiny slice of the picture. Scale that across a growing company with thousands of interactions per month, and you get prompt roulette: inconsistent answers, missed patterns, and token limits that cut you off before you get to the insight.
NEXT AI takes a fundamentally different approach:
Interpreted intelligence layer: NEXT continuously ingests, cleans, and structures raw feedback into a governed data model — so every query draws from reasoned, organized customer intelligence, not raw text dumps. This is what makes it token-efficient, fast, and reliable at scale.
Reasons across all the data: Not a sample, not a list of verbatims. NEXT quantifies themes, ranks drivers, and surfaces evidence across the full dataset — producing outputs teams can actually act on.
Autonomous distribution: Insights are pushed to Slack, backlogs, and workflows automatically — so people and AI agents act without chasing.
Competitive intelligence built in: NEXT analyzes what customers say about other BNPL providers, keeping Pledg's positioning sharp.
Privacy by design: EU-hosted, automatic PII redaction and data anonymization across all channels. Pledg co-developed new privacy capabilities with NEXT — including anonymization for support tickets and expiration policies for sensitive recordings — features now available to all NEXT customers.
Leadership lessons
Distribution beats documentation. Insights only matter if they reach the right people in time to act. Dashboards and reports sit. Push-based intelligence moves.
Actionable beats comprehensive. A list of 500 verbatims isn't an insight. Ranked drivers with evidence and clear direction — that's what teams need to make decisions.
Evidence ends debates. When every recommendation is grounded in counts, quotes, and traceable evidence, alignment happens faster and with less politics.
Speed compounds. Shorter feedback loops mean fewer mis-bets, less rework, and a roadmap that moves with confidence.
"The biggest unlock wasn't just insights — it was speed and distribution. NEXT AI turns customer feedback into updates our teams actually see and act on, automatically."
– Adrien Bonhomme, Chief Product Officer
Looking ahead
Pledg is already using NEXT AI as its Customer OS — the intelligence layer that bridges the gap between raw customer signals and the decisions that shape their product, marketing, and growth. As agentic workflows become embedded in how teams operate, Pledg is well positioned to extend that layer further: powering more autonomous feedback-to-action loops while keeping privacy and trust at the center.