MarketSignal — AI-Powered Local Competitive Intelligence

Overview

MarketSignal is an AI-powered local competitive intelligence platform built for multi-location businesses in Guadalajara, Mexico. It monitors Google Maps rankings, review velocity, rating trends, and competitor activity — then distills everything into a weekly WhatsApp report with one specific, high-impact action item.

The platform is designed around a phased go-to-market strategy. Phase 1 validates willingness to pay through a manual service model with 3-5 paying clients. The internal dashboard handles data entry, delta computation, and report generation. Phase 2 adds automated scraping and WhatsApp delivery. Phase 3 opens self-service onboarding.

The architecture reflects this phased approach: append-only snapshots allow re-analysis without data loss, JSONB columns store raw scraper output for future flexibility, and the multi-org data model ([orgSlug]) supports scaling from day one without migration.

The Challenge

• No affordable local SEO intelligence in Mexico: Enterprise tools like BrightLocal and Whitespark don't serve the Mexican market well. Pricing is in USD, support is in English, and the tools assume US-centric search patterns.
• Competitors move fast, visibility is slow: A competitor can gain 20 reviews in a week and jump 3 positions in Google Maps. Without systematic tracking, businesses don't know until their phone stops ringing.
• Insights without action are useless: Most rank tracking tools show data but don't tell you what to do. Business owners need one clear recommendation per week, not a dashboard full of charts.
• Reports need to meet clients where they are: In Mexico, WhatsApp is the default business communication channel. Email reports get ignored.

The Solution

Systematic rank and review tracking: Weekly snapshots capture Google Maps position, review count, average rating, and competitor data for every tracked keyword-location pair. Week-over-week deltas surface changes automatically.

AI-powered analysis with Claude: Raw delta data feeds into Claude API (Sonnet) to generate natural-language insights and one specific action item per location. The AI doesn't just describe what changed — it recommends what to do about it.

WhatsApp-native delivery: Reports are formatted specifically for WhatsApp readability — short paragraphs, bold key metrics, clear action items. No PDFs, no links to dashboards. The insight arrives in the conversation thread where the business owner already operates.

Internal dashboard for operational efficiency: The Next.js dashboard gives the operator (Phase 1: manual; Phase 2+: automated) full visibility into rankings, alerts, competitor comparisons, and report status across all organizations.

Technical Highlights

• Append-only snapshot architecture: Snapshot data is never updated — only new snapshots are inserted. This enables historical re-analysis and audit trails without complex versioning.
• JSONB raw data columns: Raw scraper output is stored alongside structured fields, preserving the full response for future analysis without schema changes.
• Multi-org from day one: Dynamic [orgSlug] routing and org-scoped queries avoid a painful multi-tenancy migration later.
• Report status flow: Draft → Approved → Sent → Delivered status tracking ensures no report goes out without review.
• Drizzle ORM as single source of truth: Schema-driven types flow from database to API to UI with no manual type synchronization.
• Phased architecture: The codebase is structured to support Phase 2 automation (Playwright scraper, WhatsApp API) without rewriting Phase 1 code.

Results

For the End User / Team:

• Multi-location businesses receive structured competitive intelligence they previously had no access to
• One actionable recommendation per week reduces decision fatigue
• WhatsApp delivery achieves higher read rates than email-based alternatives

Technical Demonstration:

• Full-stack Next.js 15 application with App Router, server components, and dynamic routing
• AI integration pattern: structured data → Claude API → formatted business output
• Production-grade data modeling with append-only snapshots and JSONB flexibility
• Phased product architecture that scales without rewrites