Want a SaaS idea that sells? Borrow it from 1-star reviews

When I built Jotform almost 20 years ago, I had one unfair advantage: I worked at a media company drowning in web forms.

Every week, someone asked me to code a new form from scratch, and every week I thought, this is ridiculous — why isn’t there an easier way? That frustration eventually became my startup idea.

But let’s be honest: Not everyone gets that kind of lucky break.

If I had to start from zero today — no idea, no funding, no network — I’d first pick a software category where people already spend money, then mine reviews of existing tools in that space.

And I’d automate the whole thing with an AI agent. Here’s how.

1. Pick a SaaS neighborhood

Before you dig into reviews, define the market space you’re curious about. This isn’t a forever decision — it’s just to focus your research.

Start with a noisy market (lots of tools, lots of chatter) where businesses already spend money every month — think project management, HR, analytics, CRM, etc.

Quick ways to pick:

• Start with your experience. If you’ve worked in marketing, look at SEO tools. If you’ve done HR, look at ATS software.
• Browse G2/Capterra categories. Find software with lots of reviews (100+), meaning there’s demand but probably gaps too.
• Find industries with recurring spend. B2B is ideal because budgets are bigger and churn is painful for them.

Let's use an example moving forward: customer feedback tools.

2. Gather the raw reviews

The raw material here is reviews. That’s what you’ll mine for ideas.

How to grab them:

• Go to G2 or Capterra
• Open the top few products in your category
• Copy 20–50 reviews from each into a doc
• Grab the star rating and what people wrote in “pros” and “cons”
• If copy-paste is blocked → screenshot it and move on

If you want to speed it up:

• Scrapers like Apify, BrowseAI, or Instant Data Scraper sometimes work
• When they do, export everything into a sheet (much easier to scan)
• Don’t expect them to always work — some sites block them

How many do you need?

• A hundred reviews is enough to see patterns
• 200–500 is better if you want things to really stand out

Don’t overthink it. Even a smaller batch will do the job for the next step

3. Feed reviews into AI for summarization

You don’t need to slog through hundreds of reviews line by line. Let AI do the heavy lifting.

Tools: ChatGPT, Claude, Gemini, or a notebook-style AI like NotebookLM.

Prompt template:

"You are a market research analyst. Here are 200 customer reviews for multiple products in the Customer Feedback Tools category.Summarize the top 10 pain points customers mention across all reviews.Highlight features people constantly complain about being missing, hard to use, or overpriced.Spot any patterns where customers explicitly say: 'I wish there was a tool that...'"

In minutes, you’ll have a clear, ranked list of pain points, like:

• “Integrations with Slack keep breaking.” (47 mentions)
• “Reporting dashboards are too technical.” (32 mentions)
• “Pricing is too high for startups.” (19 mentions)

💡 Pro tip: If your reviews are in a CSV, paste chunks or upload the file directly into AI tools that handle document analysis (e.g., ChatGPT with Advanced Data Analysis or Claude with file input).

4. Extract the gaps and opportunities

AI will give you a summarized list of problems. Look for:

• Repeated complaints. If dozens say “hard to integrate with Slack,” that’s a signal.
• High emotional language. Frustration = opportunity.
• Missing feature wishes. “I wish it could do X” = gap you can fill.
• High cost vs. value. “Too expensive for what it does” hints at a cheaper, lean alternative.

In our example of a feedback tool, output might be:

• “Integrations with Slack and Teams are unreliable.”
• “Analytics dashboards are confusing for non-technical users.”
• “Pricing is too high for startups, no simple tier.”
• “No quick way to send micro-surveys in-app.”

From this, you might imagine a lean MVP: A lightweight, affordable feedback tool for SaaS teams that sends quick in-app micro-surveys and integrates reliably with Slack.

5. Validate before building

Don’t start coding yet. First, test the idea’s pull.

1. Use an AI website builder (Lovable, v0, TeleportHQ, WordPress AI builder, Hostinger AI, Typedream, etc) to spin up a 1-page landing page. Going back to our example:

   • Headline: “Fast, reliable micro-surveys for SaaS teams.”
   • Subhead: “Send in-app surveys and get instant Slack notifications. Cheaper and simpler than big feedback tools.”
   • Email capture form (use Jotform or Tally).

2. Share it in relevant communities (Reddit, Indie Hackers, Twitter/X).

3. Run $50 in Google or LinkedIn ads targeting SaaS product managers.

4. Your goal: collect 20–30 emails of people interested.

This fake-door test tells you if the pain point is real enough for strangers to click and sign up.

6. Use AI to design your MVP scope

Once you see interest, go back to your AI research notes and scope the MVP.

Prompt example:

"From the customer pain points we extracted, which 3 features should be in a bare-minimum MVP that solves the biggest frustrations? Rank them by importance."

You’ll get a prioritized list to avoid feature creep. Build only the essentials. Tools like Cursor or Copilot can help you prototype faster.

7. Iterate based on feedback

• Share early builds with people who signed up on your landing page.
• Use AI to analyze their comments just like you did with reviews.
• Improve quickly based on what’s truly painful or missing.

Bottom line: You don’t need a stroke of genius. The market is already telling you what’s broken. Use AI to listen faster, cut through the noise, and build something people are quietly begging for.