Skip to content

Beta: real launches, updated every morning. Rankings may change while we tune them.

Methodology · updated 18 October 2026

How we rank.

Every launch gets one Launch Score from 0 to 100. Here is exactly how we work it out and what we never do.

The Launch Score

Each morning we read every source below and keep what is new: Show HN and Launch HN posts created during the previous full UTC day (points are read at about 07:30 UTC, so every post has had at least 7.5 hours after its day ended), GitHub stars gained since the previous morning's snapshot for repos created in the last 30 days, Hugging Face Spaces and models created in the last 60 days that are trending, apps released in the last 90 days that are in the US App Store top 100, and launch posts from company newsrooms in the last 36 hours. Launches with the same website, repo or app are one product.

Every source has one signal:

SourceSignalCounts fromStrong atWeight
Show HNPoints on Hacker News103001
Launch HNPoints on Hacker News32001
GitHubStars gained since yesterday3020000.9
Hugging FaceHugging Face trending score104000.8
App StoreChart position (101 minus the position)11000.85
NewsroomAn official launch postEvery launch postNo score of its ownBonus only
  1. Strength within the source. Half of it is the launch's percentile among that source's launches the same day, half is its signal divided by the “strong at” value, capped at 1. That is multiplied by the source's weight.
  2. The best source counts. A product's score is 100 times its strongest source.
  3. More sources add a little. Each extra source adds 6 points, up to a maximum score of 100. A company newsroom post counts only this way: it never ranks a product on its own, and posts with no measured launch are listed separately as Company news.
  4. New launches go first. A launch in the daily top 100 for the first time gets 8 extra points. A launch loses 6 points for each of the last 3 days it already spent in the top 3, so the top of the list turns over instead of showing the same products every morning.
  5. No source fills the top. One source can take at most 40% of the top 30; the rest move further down the list. We also keep only the 15 most trending Hugging Face entries each day.

The weights and “strong at” values above are first guesses; we will tune them after a week of data and log the change below. Every product page shows its own score breakdown.

A launch whose own description isn't in English is left out. When a description is under 8 words, we take the fuller text its own source gives (a GitHub README, Hugging Face model card, App Store description, Hacker News post or the site's meta description), and every morning we try again for those still short; we never write a description ourselves. Funding filings are listed under Just raised and never change a score.

Momentum

Still rising lists products first seen in the last 90 days that were in the daily top 100 on at least 3 of the last 7 days, and whose summed score over the last 7 days beats the 7 days before. Product pages chart the daily Launch Score for the last 7 days once a product has been ranked on two or more days.

What we never do

  • We never sell a place in the ranking. No paid line-skips, vote multipliers or sponsored rows.
  • A sponsor, when we have one, gets one slot outside the list, clearly labelled.
  • We never use Product Hunt data unless Product Hunt gives us written permission. We only link to a Product Hunt page when a maker adds it.

Mistakes

Every number on a product page links back to its source. When a source changes or we find a mistake, the next morning's ranking corrects it. To report one, write to hello@dailylaunch.org.

Sources and terms

We only read sources whose terms allow it, and we link back to each one.

Show HN

Show HN posts through the public Hacker News Search API by Algolia.

Counts: Points on the post, read the next morning for posts from the previous UTC day.

Terms: Public API. We link to the HN thread and never copy comments.

Launch HN

Launch HN posts by Y Combinator companies, from the same public API.

Counts: Points on the post, read the same way as Show HN.

Terms: Public API. We link to the HN thread.

GitHub

Public repositories through the GitHub REST API.

Counts: Stars gained since the previous morning; for a repo seen for the first time, its average stars per day since it was created.

Terms: GitHub's API terms. We show the repo's own README image, never private data.

Hugging Face

Public models and Spaces through the Hugging Face Hub API.

Counts: The Hub's trending score.

Terms: Public API. We link to the model or Space page.

App Store

Apple's public top-chart RSS feeds.

Counts: Chart position in the US top 100, counted as 101 minus the position.

Terms: Apple's marketing feeds. We link to the App Store page.

Uneed

Public launch pages on Uneed.

Counts: Votes on launch day.

Terms: Used only once we have confirmed their terms allow it.

Newsroom

Official launch posts and press kits on a company's own site.

Counts: No number. A newsroom post adds the extra-source bonus to a product that is ranked anyway.

Terms: We link to the post and write our own one-line summary.

SEC Form D

SEC Form D filings from EDGAR, matched to products by company name and domain.

Counts: Amount raised, shown as a badge. It never changes the score.

Terms: Public US government data.

Changes

  • 2026-10-22. Every launch page is open to search engines again, however short its own text, and so is Still rising when it is empty. Short descriptions are not hidden; the daily re-check keeps looking for fuller text from each launch's own source.
  • 2026-10-20. Correction: GitHub repos that shared one homepage were counted as one product, so some pages mixed several repos (PhotoCraft held 11 storytold repos). Each repo is now its own product, keyed by its full owner/repo, and each past day is credited to the repo that earned it. Product pages now show the product's own words once, a timeline with only each day's signal and the time the ranking was built, and rank changes relative to the day shown. Pages with under 20 words of the product's own text are kept out of search until its source gives us more, unless they are in the day's top 20.
  • 2026-10-18. A launch whose description is under 8 words now gets the fuller text from its own source: the first paragraph of its GitHub README or Hugging Face model card, its App Store description, its Hacker News post, or its site's meta description. Launches still under 8 words stay in the ranking, and every morning we look for a fuller description again.
  • 2026-10-18. The daily run starts an hour earlier, so Hacker News points are read at about 07:30 UTC instead of 08:30, after at least 7.5 hours instead of 8.5. The scoring rules are unchanged. The source list below now names the exact signal each source is scored on, and the chart on product pages is labelled as the daily Launch Score it shows.
  • 2026-10-17. Launches ranked for the first time now get 8 extra points, and a launch loses 6 points for each of the last 3 days it spent in the top 3. Products that stay strong on their source still rank, but the top of the list now changes from day to day.
  • 2026-10-15. Show HN and Launch HN now count posts from the whole previous UTC day, so every post has the same time to collect points. Newsroom posts no longer have a score of their own: they add the extra-source bonus to a product that is ranked anyway, and newsroom-only posts are listed separately as Company news. Every product page shows how its score was worked out.