---
name: competitor-feature-matrix
description: "Turn a list of competitor URLs into a normalized feature and pricing matrix you can paste into a deck — without the 'plan names mean different things at each company' problem."
title: Competitor Feature Matrix
category: research-analysis
difficulty: beginner
author: admin
icon: 📊
input: text
output: structured-json
phase: transform
domain: research
tags: competitive-analysis,web-scraping,feature-extraction,pricing-intelligence,matrix-generation,data-normalization,embedding-similarity,semantic-aliasing,csv-export,provenance-tracking
best_for:
  - Product strategy and competitive positioning
  - Sales enablement decks and battle cards
  - Market research and feature benchmarking
  - Go-to-market analysis across SaaS competitors
---

## Description

Given 5-20 competitor URLs, crawls each site's pricing and feature pages and emits a CSV where every row is a feature and every column is a competitor. Plans are normalized to tiers (free / paid / enterprise) rather than vendor-specific names. Missing data is explicitly marked 'not disclosed' instead of left blank.

## Why it works

Feature matrices usually fail because each vendor uses different words for the same thing and each reader interprets blanks differently. Forcing a canonical feature list up front and distinguishing 'absent' from 'undisclosed' prevents the matrix from silently lying.

## How it works

1. Scrape each pricing page + features page into clean markdown. 2. Ask the LLM to extract features as (name, description) tuples per vendor. 3. Merge synonymous features across vendors using embedding similarity + confirmation prompt. 4. For each (feature, vendor) cell, re-prompt against just that vendor's source markdown with three options: present / absent / undisclosed. 5. Render as CSV with a provenance column linking to the page each claim came from.