Customer Feedback Search Engine

❓ Problem

Customer feedback isn't missing. It's scattered. The average software company now hears from customers across 15+ separate surfaces: Zendesk and Intercom tickets, Gong sales calls, App Store and Play Store reviews, Reddit threads, Discord servers, X replies, in-app surveys, NPS comments, community forums. The signal is all there, written down, in customers' own words. Nobody can see it in one place.

So watch what happens. A PM commissions a research project that takes three weeks and is half-stale the day it lands. An engineer ships against whichever ticket shouted loudest this morning. And the CEO forms their entire worldview of "what customers want" from the same five tweets that happened to cross their timeline. Three people, three completely different versions of customer truth, none of them right. As one VoC founder put it, teams are left painstakingly parsing disconnected streams by hand while crucial feedback falls through the cracks and quietly fuels churn.

The real kicker? This was never a data problem. It's an access problem. The feedback already exists, fully articulated, sitting in a dozen tools behind fifteen logins and twelve different search boxes. There's just no shared brain. Everyone in the building is making customer decisions from the tiny sliver they personally happened to read. You don't know what your customers think. You know what the last person told you.

✅ Solution

Murmur (murmur.ai) is a managed pipeline that ingests every customer-facing channel, normalises and semantically indexes the entire corpus, and exposes it via MCP to whatever LLM the company already uses (Claude, ChatGPT, Cursor, an internal LibreChat).

• One pipeline, every channel. Connect the sources once (support, calls, reviews, social, surveys, community) and Murmur keeps an always-fresh, deduplicated, semantically indexed corpus of everything customers have ever said, across every surface.
• Answers, not dashboards. Instead of a chart to interpret, you ask a question in plain English from the chat tool already open on your screen: "are merchants complaining about the new checkout flow?" Back comes a one-line verdict in 30 seconds, grounded in the actual tickets, threads and calls, with every claim cited to its source.
• Lives where the work happens. Because it speaks MCP, Murmur appears inside Claude, ChatGPT, Cursor or an internal chat tool. No new tab, no new login, no asking the whole org to learn yet another piece of software.
• Trust through receipts. Every answer links straight to the verbatim feedback behind it, so a PM can click from "sentiment is sliding on mobile" through to the eleven exact reviews that prove it.

The wedge is narrow on purpose: give one team (usually product or CX) a research analyst that never sleeps. The corpus underneath quietly becomes the company's single source of customer truth, and it spreads sideways as engineers and execs start asking it questions without being told to.

📊 Key Numbers

Market size

• The Voice of the Customer software market sits at roughly $16.2bn in 2025 and is forecast to reach ~$52bn by 2035, growing at about 15.9% a year. The broader customer-analytics market is ~$13bn in 2025 heading toward ~$29bn by 2029. That is the pool Murmur swims in, and it is compounding in the mid-teens.
• SAM: the realistic serviceable market is software and digital-first companies with enough multi-channel feedback to need this and enough budget to buy an AI-native tool. Call it ~50,000 Series A-and-up companies globally. At a blended ~$50k ARR that is a ~$2.5bn serviceable slice.
• SOM: landing 500 mid-market accounts at ~$60k ARR is ~$30M ARR, before any expansion. That is a focused, winnable early target inside a huge category.

ARR potential