I Fed Our Competitor's G2 Reviews to an AI. It Wrote Their Next PRD.

You have a new mandate: "Get a better handle on the competition." So you open up G2, Capterra, and a dozen browser tabs. You start copying and pasting user reviews into a spreadsheet. "Slow loading times," you type into a cell. "Confusing UI for new users," goes another.

Three hours later, your spreadsheet has 200 rows. Your eyes are glazing over. You have a list of complaints, but no real insight. You’ve found individual trees, but the shape of the forest is a complete mystery. You've spent half a day on manual labor that a machine could do in three minutes.

This isn't competitive analysis. It's data entry.

The patterns—the deep, systemic frustrations that signal a true market opportunity—are buried in the volume. Your brain, prone to recency bias, will latch onto the last three reviews you read. The real gold is the slow-burn problem mentioned consistently for two years straight. You won't find that with a spreadsheet.

Let's try a different way.

Step 1: Get the Raw Material

First, you need the reviews. All of them. Don't just read the first two pages. Go deep. Use a scraping tool like Apify or Browse AI to pull every single review for your top two competitors into a text file or spreadsheet. We're looking for volume. Five hundred reviews is better than fifty. A thousand is better than five hundred. This is the raw ore we're about to refine.

Step 2: Write a Prompt That Thinks Like a PM

Dumping 1,000 reviews into an AI with the prompt "summarize this" will give you a useless word cloud. You need to give the AI a role, a context, and a very specific job to be done. You need to write a prompt like it’s a product spec.

Here’s a template to start with:

ROLE: You are a Principal Product Manager at a B2B SaaS company. You are an expert at synthesizing qualitative customer feedback and identifying strategic product opportunities from competitive weaknesses.

CONTEXT: I am providing you with 750 user reviews for our competitor, [Competitor Name]. Our product is [Your Product Name], and we help [Your Target Persona] solve [Core Problem]. [Competitor Name] is our main rival in this space.

TASK: Analyze the attached reviews and perform the following actions:

1. Identify Core Jobs to be Done: Based on the positive reviews, what are the top 3-5 "jobs" that customers are hiring this product to do?
2. Cluster Critical Pain Points: Analyze the negative and mixed reviews. Identify the top 5 most frequently cited pain points, frustrations, or feature gaps. Label each cluster clearly (e.g., "Onboarding Complexity," "Integration Failures," "Reporting Inflexibility").
3. Provide Verbatim Evidence: For each pain point cluster, provide 3-5 direct quotes from the reviews that best exemplify the problem. Do not summarize the quotes.
4. Draft an Opportunity PRD: Based on the most severe and frequently mentioned pain point, write a one-page Product Requirements Document (PRD) for a new feature for OUR product. This feature should directly solve that core problem for the competitor's unhappy users. The PRD should include:
   • Problem Statement: `A clear, concise summary of the user