The Lazy PM's Guide to Customer Research: "Review Mining" Will Save Your Ass

How to extract customer insights without actually talking to humans

The Product Manager's Dilemma

We're all drowning in a sea of bullshit. Every product manager I know is perpetually caught in the corporate vortex of "just one more deck," while the actual job of understanding the customer gets pushed to the mythical "when I have time" zone (spoiler alert: that time never comes). You wake up each morning with grand visions of talking to real humans who use your product, only to find yourself in meeting #7 of the day, pondering an alt career path.

So how do you maximize the little time you have leftover after playing corporate politics and justifying obvious decisions with unneccesary decks? How the fuck do you know what customers want when you haven't spoken to one in three months?

Review Mining

Enter the savior of time-strapped product managers everywhere: ChatGPT's "Deep Research" feature paired with a whole lotta customer reviews. This beautiful hack is practically corporate theft, giving you insights that would normally cost weeks of research and a hefty user testing budget.

Here’s how it works:

1. Gather the Raw Material: Dump as many customer reviews as you can find into a spreadsheet. Can't be bothered with that? Just point ChatGPT to the internet and let it do the dirty work.
   

2. Craft Your Research Prompt: Ask ChatGPT to write you a research prompt to which you can give Deep Research. Give it plenty of context, see below for how I asked ChatGPT to build me a prompt to better understand customers via reading through 30k plus reviews.
   
   Please conduct a comprehensive deep-dive analysis on customer reviews of the PODS moving and storage service. Utilize the provided CSV file as your primary resource, supplemented by any additional reviews available online.
   
   Extract insights clearly categorized into the following areas, supported by specific, illustrative quotations:
   
   Triggers: Identify specific scenarios or events that prompted customers to start their buying journey or recognize they had a need. Example: "We had to move quickly due to unexpected job relocation."
   
   Goals: Clearly define the customer's functional objectives or desires. Example: "I wanted an easy, stress-free moving experience."
   
   Pain Points: Document specific challenges or issues customers experienced with their previous situations or alternative solutions. Example: "Previous moves were always stressful and costly."
   
   Other Solutions: Detail any alternative products, services, or methods customers considered, tried, or used prior to PODS. Example: "Initially, we thought about using a traditional moving truck or a local moving company."
   
   Anxiety: Identify customer concerns or hesitations that might prevent or delay the decision to use PODS. Example: "I was worried my belongings might get damaged or lost."
   
   Selfish Desire: Highlight deeper emotional or social motivations behind the purchase. Example: "I wanted peace of mind knowing my family's belongings were safe."
   
   Swipeable Language: Capture particularly effective phrases or language used by customers in their reviews, suitable for direct use in marketing.
   
   Key Insights: Synthesize the most critical learnings from the review mining exercise to inform positioning, messaging, and future strategic direction. Example: "PODS uniquely solves last-minute moving challenges and significantly reduces stress compared to traditional options." Include direct quotations from reviews that clearly support each finding. If the provided CSV does not sufficiently address a category, perform supplemental research by consulting additional online sources such as Google reviews, Trustpilot, Yelp, and other relevant sites for PODS reviews.
   

3. Turn on Deep Research mode: Start a new chat, flip on Deep Research mode, and copy/paste your prompt.

   

4. Take a Break: Pour yourself a stiff drink or a quadruple espresso. You've earned it by outsourcing your job to an AI.
   

5. Harvest the Gold: After 10-15 minutes, ChatGPT will spit out enough customer insights to make your research team blush with shame at how quickly you've replicated their quarterly output.

This shit actually works. I tried it with PODS moving service and got back insights so specific, it was like I'd painstakingly interviewed 50 customers and then meticulously distilled my findings into an actionable guide.

The AI-Enhanced PM

And just like that, you can stroll into meetings armed with actual customer language. Your roadmap decisions are backed by real human desires. Your feature prioritization has at least a plausible shield against the exec who won't shut the fuck up about their pet project.

You've achieved what seemed impossible: deep customer insights without canceling happy hour to hear Karen from Oklahoma describe her user journey – you've transcended such primitive research methods. Now go forth and product, you magnificent, time-hacking bastards.