Find Sales Prospects Using Google Reviews with Codex

Reviews tell you what’s broken. A dentist with “couldn’t get an appointment for 3 weeks” reviews needs better scheduling. A gym with “app never works” complaints needs a tech partner. Codex connected to AgentPatch can search Google Maps for businesses, analyze their reviews, and build a prospect list based on real customer complaints.

Why This Matters

The best cold outreach references a specific problem the prospect actually has. Review data gives you that for free — customers have already written down what’s wrong. The issue is scale: reading reviews for 20 businesses by hand takes time.

Codex can search Maps, pull review data via Google Search, identify complaint patterns, and output a ranked prospect list — all from the terminal.

Setup

Add AgentPatch to ~/.codex/config.toml:

[mcp_servers.agentpatch]
url = "https://agentpatch.ai/mcp"
bearer_token_env_var = "AGENTPATCH_API_KEY"

Then set your API key:

export AGENTPATCH_API_KEY=your_api_key

Replace your_api_key with your actual key from the AgentPatch dashboard. Codex discovers all AgentPatch tools automatically on next start.

The Workflow

Step 1: Find Businesses

Tell Codex what category and location to search:

“Search Google Maps for auto repair shops in Nashville. Return 10 results with names, ratings, and review counts.”

Codex calls the Maps tool and returns the list.

Step 2: Mine Reviews for Signals

“For each shop, search Google for their reviews. Identify recurring complaints — especially ones related to communication, pricing transparency, or wait times. Summarize the top pattern for each.”

Codex searches for reviews of each business and extracts the common threads.

Step 3: Rank and Output

“Which shops have communication or transparency complaints that a CRM or customer portal could address? Rank by signal strength and output as structured data.”

Review-based prospect list — auto repair shops with communication issues:

1. Music City Auto (3.4★, 198 reviews)
   Pattern: 23 reviews mention "never called me back" or "had to call 3 times for updates"
   Signal: Strong. Core complaint is communication during service.

2. Broadway Garage (3.7★, 91 reviews)
   Pattern: 11 reviews mention surprise pricing, "bill was way more than the estimate"
   Signal: Strong. Pricing transparency issue — a customer portal with estimate
   tracking would address this.

3. Hillsboro Auto Care (3.9★, 56 reviews)
   Pattern: 6 reviews mention slow response to appointment requests
   Signal: Moderate. Likely manual scheduling — could benefit from automation.

Remaining 7 shops: No strong communication-related patterns. Complaints were
mostly about quality or pricing (not transparency).

You have three prospects with specific, defensible reasons to reach out.

Wrapping Up

Review data is the most underused prospecting signal available. Customers tell you exactly what’s broken — you match that to what you sell. Codex with AgentPatch handles the research pipeline: Maps for discovery, Search for review analysis. Try it at agentpatch.ai.

← Back to blog