ReviewIQ

LONG STORY SHORT

I built the tool I always wished existed as a PM

Every product manager I know has spent a Friday afternoon drowning in G2 reviews, Reddit threads, and App Store screenshots — trying to figure out what users actually hate about the competition. It takes hours. It's messy. Half the time you don't even end up using it because the deadline already passed. So I built ReviewIQ. Paste in competitor reviews, get back a structured breakdown of unmet needs, recurring complaints, feature gaps, and an opportunity score — in under 30 seconds.

THE PROBLEM

Competitive research is broken, and everyone just accepts it

The signal is already out there. Thousands of users are openly complaining about your competitors on G2, Reddit, the App Store. But nobody has time to read all of it, synthesize it, and turn it into something useful for a roadmap conversation. The result? Junior PMs skip it entirely. Senior PMs do a rushed version under time pressure. Everyone is making product decisions with incomplete competitive context.

"Competitive research typically takes 2–4 hours per competitor. It's inconsistent, skewed by recency bias, and rarely documented in any shareable format."

THE USER

PMs who know the insight is there — they just can't afford the time to find it

I wasn't designing for everyone. I was designing for one person: the PM at a 40-person SaaS startup who knows competitive research matters but keeps deprioritizing it because it takes half a day.

THE DECISION

I had three options. One was obviously right.

Before building anything, I mapped out the real choices.

THE PRODUCT

Four features. Everything else was cut.

One filter: does this get someone from paste to insight in under 30 seconds?

• F1

  Review Input

  Any format. G2, Reddit, App Store. 200 char min, 5K char max.

• F2

  AI Analysis Engine

  Claude API extracts Unmet Needs, Complaints, Feature Gaps, and Opportunity Score (1–10).

• F3

  Shareable Output Card

  Clean results card. One-click copy. PNG export for LinkedIn and Slack. Unique URL per report.

• F4

  Report History

  Every report persists anonymously at a unique URL. No login required.

TRADEOFFS

Every choice was a deliberate bet.

• MANUAL INPUT VS AUTO-SCRAPING

  Manual paste adds friction — but eliminated weeks of scraper engineering, ToS risk, and rate-limiting complexity. The real test was whether the AI output was valuable enough that people would endure the paste step. They did.

• ANONYMOUS SESSIONS VS USER ACCOUNTS

  Removed the biggest conversion killer on day one. The v1 goal was usage, not retention. Unique URLs gave users enough to work with.

• 5K CHARACTER CAP

  Controlled API costs, kept analysis under 10 seconds. Chunking for larger inputs scoped to v2, documented in the PRD.

• SPEED VS POLISH

  Shipped in one weekend using Lovable components. The value proposition was the AI output, not the interface. Polish is a v2 priority.

IMPACT

It worked.

LEARNINGS

What this taught me about building and shipping.

• Writing the PRD first changed everything.

  I wrote a full spec before touching any code. When tradeoffs came up during the build, I already knew what the product was trying to do. The decisions were fast.

• Constraints are actually a gift.

  One weekend forced brutal prioritization. I had to identify the single thing that had to be true for the product to work. For ReviewIQ, that was: does the AI output feel genuinely useful? Everything else was negotiable.

• Shareable output is a growth strategy, not a feature.

  The PNG export became the main distribution channel. People shared ReviewIQ output cards on LinkedIn and that drove traffic. Build for shareability from day one.