Over the past two days I checked 90 browser-extension ideas against real store numbers before building anything: Chrome Web Store and AMO user counts, ratings, prices, and what the browser already does built in. My gate was simple — no build without two kinds of hard evidence: a real incumbent's traction numbers, and a paid version of something similar with visible signs that someone pays for it. 86 of the 90 failed the gate. Four passed and got built. None of the four has made a sale yet, so this is a report on the gate, not a victory lap.

Here are the five patterns that killed almost everything.

1. A free incumbent already owns the category

Dark Reader: ~7M users on Chrome, 1.37M on Firefox, 2.7M on Edge — open source and free. Read Aloud: 6M users, free and open source. SponsorBlock: ~2M on Chrome and 730K+ on Firefox, free and open source, with a 13M+ crowdsourced timestamp database a new entrant cannot cold-start. Video DownloadHelper: ~6M users. Immersive Translate: 10M+ users. Even a supposedly dead product wasn't dead: MarkDownload still has 11,307 daily active users on AMO at 4.85 stars. When the incumbent is free, maintained, and has millions of users, "I'll build a better one" is not a demand signal.

2. The browser already absorbed it

Chrome's built-in Memory Saver absorbed tab suspenders. Chrome 145 (early 2026) added native Split View, so two tabs side by side no longer needs an extension at all — Edge and Firefox had it built in for years. Chrome's built-in password manager, translate, tab search, and per-site zoom each absorbed a whole category. Firefox has had fingerprinting protection built in since version 67. If the feature is one browser release away from being a checkbox, there is no product.

3. The only paid version is a subscription

Toby charges $6/month. Text Blaze's paid tiers are subscriptions. Immersive Translate Pro is $14.99/month. Raindrop Pro is ~$28/year. axe DevTools Pro is sold through Deque's sales team. These are all real paid products — but every one of them is a subscription or an enterprise sale, because the thing being sold needs ongoing work or ongoing cost. My model is a one-time ¥19.9 activation code. A category where the only proven willingness to pay is a subscription is not evidence for a one-time purchase.

4. The paid feature needs a server

Distill's local checks are free and unlimited; it charges from $15/month for cloud checks — frequency and quota from a server. Price-history extensions need crawling infrastructure. Translation features need API spend. Cloud sync is what Super Session Manager charges for, while the local part is free. A one-time payment cannot fund a recurring server bill, so any idea whose paid tier is "we run something for you every day" fails my model even when users clearly pay for it.

5. The niche is too small to pay for itself

Linkclump Pro has ~20,000 users and is entirely free. A text expander that launched on 2026-11-10 had 0 votes. An audio-recorder extension has 361 users. A video downloader has 817. Some categories have real users and zero paid evidence of any kind. A few thousand users who have never paid anyone for this is not a market — it's a hobby.

What passed the gate

Four ideas cleared it, and I built all four:

  • ConvoExport — export a ChatGPT or Claude conversation. The incumbent AI Toolbox has 40,000+ users at 4.7/5 and charges $9.99/month per module; the one-time alternative didn't exist.
  • YT Transcript Export — single-video transcript export, with playlist-batch export as the paid tier. No extension with over ~1,000 users did playlist-batch export.
  • FocusBlock — a local-first website blocker. BlockSite has ~1M users, but its July 2026 privacy policy says it collects browsing data and shares it with affiliates; FocusBlock stores everything locally and makes zero network requests. It is live on Edge now: https://microsoftedge.microsoft.com/addons/detail/focusblock/kdpeddhiegpfadpjiaieoaeiloebgdhp
  • SessionSnap — a cross-browser session manager. Session Buddy has 1M users and a 3.3-star rating that fell from 4.7, hasn't been updated since 2026-04-28, and has no official Firefox or Edge version — I verified the AMO listing 404s and the official site only mentions the Chrome store.

The honest footnote

The gate is good at killing bad ideas. It has not yet proven it can pick winners: four built, zero sales so far, one live on Edge, three still in Edge review. The next test is whether anyone installs the live one. I'll report that number when it moves — including if it stays at zero.

The gate, as a checklist

  • Name the incumbent, with its real user count from the store page.
  • Name the paid version, with its real price and some sign someone pays it.
  • Check what the browser already does built in.
  • Check whether the paid feature needs a server.
  • If all four answers point at free, built-in, subscription, or server — stop.

Note: the numbers above are store snapshots from checks on 2026-11-15–17. "Users" means the store's published user count, not installs I measured.