<p><b>Startup description</b></p><p>My extension adds an "Improve" button next to the prompt box on ChatGPT, Claude, Gemini, Lovable and Base44. One click rewrites a rough prompt in place — no copy-paste, no tab switching. A second mode, Craft, builds a prompt from a few guided questions.</p><p>We started it in 2024 because the three of us were rewriting our own prompts two or three times before getting a usable answer. It has been live for two years, is Manifest V3 compliant, and carries Google's Featured badge with 76 ratings at 4.6 stars.</p><p><b>Key highlights:<b></b></b></p><ul> <li>Annual revenue: $4,141 TTM, against $2,534 the prior year</li> </ul> <ul> <li>Annual profit: approx. $4,021 — running costs are about $10/month</li> </ul> <ul> <li>Business margin: approx. 97%</li> </ul> <ul> <li>Churn rate: 6% monthly, improved from 20% a year ago</li> </ul> <ul> <li>LTV: approx. $65</li> </ul> <ul> <li>CAC: approx. $0 — every install is organic Chrome Web Store search, no paid acquisition ever</li> </ul> <ul> <li>Chrome Web Store users: 7,000, across 128 countries</li> </ul> <ul> <li>Rating: 4.6 stars from 76 ratings, plus Google's Featured badge</li> </ul> <ul> <li>Subscribers: 52 active, from 258 sold lifetime</li> </ul> <ul> <li>Paywall-to-purchase conversion: 2.6%, at market for a freemium extension.</li> </ul> <p><b>Team</b></p><p>Three co-founders, all part-time alongside other work. No employees, no contractors, no salary costs.</p><p><b>Tech Stack</b></p><ul> <li>Extension: React, TypeScript, Manifest V3, built with Plasmo.&nbsp;</li> <li>Backend: Node.js API with PostgreSQL.&nbsp;</li> <li>Model providers: OpenAI and Groq.&nbsp;</li> <li>Payments: Lemon Squeezy as merchant of record.&nbsp;</li> <li>Analytics: Mixpanel, two years of history.&nbsp;</li> <li>Marketing site.</li> </ul> <p><b>Marketing and growth</b></p><p>Effectively none — which is either the weakness or the opportunity, depending on your view. All 7,000 installs came from Chrome Web Store search and word of mouth. No paid acquisition, no content marketing, no email list, no social presence.</p><p><b>Revenue and profit</b></p><ul> <li>TTM revenue $4,141 against $2,534 the prior year.&nbsp;</li> <li>Current MRR $212, down from $287.&nbsp;</li> <li>52 active subscribers from 258 sold lifetime.&nbsp;</li> <li>Monthly churn 6%, down from 20%.&nbsp;</li> <li>Seven refunds in the year.&nbsp;</li> <li>Running costs are roughly $10/month, so this is close to a pure-margin business.</li> </ul> <ul> <li>Pricing: $4.99/month, $11.99/quarter, $39.99/year. A single tier. The category norm is $9-13/month, so the price has never been tested upward.</li> </ul> <p>Revenue rose year on year, but the mix matters: new order revenue halved while renewals grew. This year was paid for largely by people who bought last year.</p><p><b>Return on investment</b></p><p>At $12,000 against roughly $4,021 of annual profit, simple payback is about three years with no changes at all.</p><p>The faster route is maintenance. On 22 July 2026 we shipped one release that repaired two broken integrations. No marketing, no new features, no price change. Active users rose 66% within three weeks and were still 50% above baseline five weeks later. Three integrations are still broken — scoped, known work with an already-measured return.</p><p><b>Startup assets<b></b></b></p><ul> <li>Chrome Web Store item with 7,000 users, transferred through Google's official process — extension ID unchanged, existing users undisturbed</li> </ul> <ul> <li>4.6-star rating and all 76 reviews, which transfer with the item</li> </ul> <ul> <li>Google Featured badge</li> </ul> <ul> <li>Full source code: extension and backend API</li> </ul> <ul> <li>Prompt engineering IP — the optimization prompts are the actual product</li> </ul> <ul> <li>.com domain and marketing site</li> </ul> <ul> <li>Two years of Mixpanel analytics history</li> </ul> <ul> <li>52 active subscriptions</li> </ul> <ul> <li>30 days of post-sale support</li> </ul> <p><b>Risks</b></p><p>Stated plainly, because a buyer will find all of this anyway.</p><p>Reliability is the main one. 43% of Improve clicks returned nothing over the year. The extension finds the prompt box by DOM selector, and those break whenever OpenAI, Anthropic or Google change their markup. Current failure rates: ChatGPT approx. 40%, Gemini 60%, Claude approx. 46%, Lovable 100%.</p><p>No failure telemetry, which is precisely why the above went unnoticed — Claude was fully broken for twelve months before anyone spotted it, and Gemini's button silently stopped rendering for five weeks.</p><p>Declining revenue. MRR is down 26% year-on-year, and new orders down 47%. The cause is the reliability problem, not demand: reach held steady while the share of users who got a working result halved.</p><p>Server-side entitlement checking is currently disabled — the API does not verify subscriptions. Needs implementing.</p><p>Mitigation for all of it: this is known, scoped maintenance work, and the July release is the evidence that doing it works.</p><p><b>Summary</b></p><p>You are buying an established distribution position with a maintenance problem — not a broken product without an audience.</p><p>7,000 users, a Featured badge, and 76 reviews at 4.6 stars take years to accumulate and cannot be bought. Reach is intact. Conversion is normal. What the product loses is the nine in ten users who click once, get nothing back, and never return.</p><p>One release in July proved that fixing it lifts usage immediately and durably, with no marketing spend. Three integrations are still broken, and we no longer have the time to maintain them. For an operator who will, the work is scoped, and the return is already measured.</p><p>✅ $4,141 TTM revenue</p><p>✅ 52 customers</p><p>✅ Business model: Freemium</p><p>✅ Built with Node.js, JavaScript, PostgreSQL</p>