· 4 min read · founder · philosophy

Why I built Fareduck

Google Flights shows you fares. Fareduck shows you the ones a human would actually be excited to see — and why that's a different problem worth solving.

By Fareduck

I've used Google Flights since the day it launched. I've used Kayak, Skyscanner, Hopper, Going (née Scott's Cheap Flights), and a half-dozen one-off newsletter experiments along the way. None of them solved the problem I actually had.

The problem isn't "show me cheap flights." Cheap flights are everywhere. The problem is show me the small handful of flights I'd actually be excited to take.

Those are different jobs.

The cheap-flight problem is solved. The "deal worth taking" problem isn't.

Open Google Flights right now. Type your home airport. Hit "Explore." You'll see thousands of fares. Most of them are noise — tiny European discount carriers to airports you can't pronounce, red-eyes with two stops in cities you wouldn't choose, business-class fares no one wants to pay for.

A deal is different from a low price. A deal has three things at once:

  1. The price is meaningfully below typical for that route.
  2. The route is somewhere you'd actually want to go.
  3. The dates work for your shape of trip.

Most existing tools optimize for #1 and leave #2 and #3 to you. Going does a great job, but it's broad — they push the same handful of routes to everyone on the list. The premium tier helps; it doesn't personalise enough.

I wanted something that knew I live in Denver, that I'd flown to Lisbon last year and loved it, that I'd rather hit Tokyo in October than August, and that I refuse to take Spirit even when it's $200 cheaper. Then surface 5–8 flights a week that match.

That tool didn't exist. So I built it.

What Fareduck does, specifically

Every day at 02:00 UTC, Fareduck runs a search for every active user. The search isn't a blunt "show me everything from JFK." It's bounded by:

Out of that search, we rank the offers with both a heuristic scorer (deterministic, predictable, fast) and a Gemini-powered ranker (catches the nuance you can't put in rules). Both are walled off from anything monetization-related. We never know which links pay us until after the ranking is done.

On Sunday morning, we send a digest with the top 5–8 picks. Most weeks it's a small list. Some weeks it's empty — better than padding with crap.

That's it. There's no "explore everywhere" button, no fare alerts I'd have to maintain, no booking engine. Just a small batch of well-curated picks weekly.

What Fareduck isn't

A few things I deliberately won't ship:

The honest bit

Fareduck makes money in three ways:

  1. Affiliate commissions — when you click "Search this fare" we route through a partner link. If you book, the partner pays us. This is most of our revenue.
  2. Plus subscriptions ($7/mo, $70/yr — launching when we hit ~100 active users) — ad-free, daily digest, target-price alerts.
  3. Ads — turned off until ~500 active users, and only direct sponsorships at first. No programmatic networks.

Ranking is walled off from monetization. That sentence is the whole point of the trust-first thing. The deal we recommend first isn't the one that pays us the most. The architecture is set up so it can't be — the ranking module literally can't import the affiliate module. If you've used enough of the "best deals!" coupon farms on the internet, you know why that matters.

What's next

I'm launching with 50 friends and family in the next few weeks. They tell me what's broken. I fix it. Then I open to a wider waitlist.

If you're reading this because someone forwarded you the URL: thanks for taking a look. Sign up here. The free tier is fully functional and I don't ask for a credit card.

If you're reading this because you found us on Google: same offer. Sign up takes about two minutes, your first digest lands within a week, and unsubscribing is one click.

— The Fareduck team