How this directory data is collected

1. Where the list comes from

The starting list was compiled from publicly published directory lists and from our own curation, then treated as candidates rather than facts. Nothing was carried over on trust: every entry had to pass the uptime check below, and every Domain Rating was re-measured from scratch.

27 entries are hand-picked and hand-written. Those carry a description written for this site, a pricing note, and a screenshot. The rest are listed with measured data only, which is why their pages lead with facts rather than opinion.

2. Uptime check

Every candidate URL was requested directly, with a HEAD request first and a full GET as a fallback for hosts that reject HEAD. Of 528 candidates, 440 responded normally and 58 returned a bot-blocking status such as 403 or 429 while still being reachable. Those 58 were kept.

25 were dead and were removed rather than left in to pad the count. That is the single clearest difference between this list and the ones it was checked against, several of which still list those same dead domains.

3. Domain Rating

Domain Rating comes from the Ahrefs free Domain Rating API, on the standard 0–100 scale. Every domain on this site is queried individually and the figure is dated, not carried over from another directory list. Because it is the same metric most SEO tools quote, the numbers here are directly comparable with what you already use.

Two limits worth stating plainly. Ahrefs returns DR for the root domain, so the handful of entries that live on a subdomain carry their parent domain’s figure and will read higher than the subdomain earns on its own. And DR describes a link profile, nothing else — it says nothing about whether a directory is still moderated, still accepting submissions, or capable of sending you a single visitor.

DR 60+

148

DR 40-59

151

DR 20-39

123

Below DR 20

86

4. Dofollow and nofollow

A dofollow link passes authority to your site; a nofollow one does not. Both can send traffic, but only dofollow listings help your own pages rank. Of 508 directories, 315 were recorded as dofollow, 186 as nofollow, and 7 are marked unverified because the link type was not confirmed.

Those 7 are labelled honestly rather than guessed. Link type also changes without notice, so treat every value here as true at the time of checking and confirm it once your listing is live.

5. Categories

Each directory sits in one of five categories. The 27 curated entries are categorised by hand, because the source taxonomy misfiled several of them.

• Launch platforms — 73 entries
• Startup profiles — 23 entries
• SaaS & software directories — 262 entries
• AI tool directories — 127 entries
• Startup communities — 23 entries

One naming point worth flagging: the Startup profiles category holds company-profile sites such as Crunchbase and Wellfound, which is a narrower set than the startup directories page, which also covers launch platforms and founder communities. Same words, deliberately different scope.

6. What this data does not tell you

A high Domain Rating does not mean a directory will send you customers. It measures backlink authority, nothing else. A DR 70 directory in the wrong category is worth less to you than a DR 25 one your buyers actually read.

Nothing here reflects approval rates, review speed, or whether a directory currently charges. Those change constantly and are set by each directory, not by this site. Uptime was checked at a point in time; sites go dark between updates.

7. Using the data

The full dataset is published in machine-readable form under a CC BY 4.0 licence, so you can check any figure on this site rather than taking it on trust.

• directories.json — every entry with its measured fields and a documented field map
• llms.txt — an index for AI agents, with the field definitions spelled out
• Any page is also available as Markdown at the same path with a .md suffix

8. Corrections

If a figure here is wrong, it is worth telling us: a directory that has gone dead, changed its link type, or started charging affects everyone reading this list. Open an issue on GitHub and it will be corrected on the next data run.