Original data

I submitted 18 sites to 235 AI directories

2,700 attempts, 81 accepted, 8 directories that ever said yes. Here is the whole dataset, including the parts that make directory submission look like a bad deal.

Data collected 2026-11-16 · CC BY 4.0 · download the CSV and JSON

The numbers first

MeasureValue
Directories probed with a real browser235
Accept a fully automated submission46 (19.6%)
Single-input or field-less pages146 (62.1%)
Require an account before the form34
Behind a CAPTCHA or challenge4
Submission attempts made2,700
Attempts a directory accepted81 (3.0%)
Directories that ever accepted anything8 (3.4%)

Read the last two rows together. 3.4% of the directory pool absorbed 100% of the acceptances. The other 227 directories took our form, returned HTTP 200, and produced nothing.

Method, so you can argue with it

Every directory was loaded in headless Chrome, not fetched as HTML. A static fetch is enough for half of them and actively misleading for the rest: single-page directories render the form after JavaScript runs, and a plain GET returns a shell with a search box in it.

What counted as automatable

A target was marked automatable only when the rendered page exposed at least two visible, fillable fields, at least one of them matched a known submission field (URL, name, email, description), and no login redirect or CAPTCHA appeared. Anything else was recorded as manual, with the reason: login wall, captcha, too few fields, or no form.

What counted as accepted

An attempt counts as accepted when the response page contained a receipt: a thank-you message, a pending-review notice, or an email-verification prompt. HTTP 200 alone counts as unknown, never as success. This is the single most common way directory-submission statistics get inflated: a 200 on a client-side form usually means the request went nowhere.

What was not measured

Whether the listing went live. At the time of writing every submission was under review, and no listing had appeared yet. Acceptances here are promises, not backlinks.

Where the pool breaks down

CategoryDirectoriesAutomatableRate
AI tool directories1592918%
Software / SaaS441227%
General web directories26519%
Local / business listings500%
Launch communities100%

Three failure modes, in order of size

One-field pages. 62.1% of the pool renders a single input, usually a newsletter box or a site search. The submission form is behind a "Submit your tool" button, or it does not exist and the site is an aggregation of products it found itself.

Accounts. 34 directories redirect to a signup page first. Most of those now use magic links or OAuth, which rules out scripted registration entirely: the login needs an inbox click, not a password.

Receipts that do not exist. A meaningful share of the SPA directories accept the POST and then change the DOM without navigating. There is no success URL, no status code, and no way for an outside observer to tell a stored submission from a dropped one.

The directories that actually accepted

Ordered by accepted submissions. The full 235-row table, with field counts, login walls and rejection reasons, is in the dataset.

DirectoryCategoryAcceptedAttemptsAutomated
www.launchingnext.comSoftware / SaaS1819yes
yaatd.comAI tool directories1818yes
onepagelove.comSoftware / SaaS1718yes
thatsmy.aiAI tool directories1618yes
aimatch.proAI tool directories818yes
insidr.aiAI tool directories218no
easywithai.comSoftware / SaaS118yes
opensaas.directoryAI tool directories118yes

What this means if you are about to do the same

A directory link is worth having when the directory has traffic, an editorial filter, or a page that ranks for its own category. Most of this pool has none of the three: the pages are nofollow, unranked, and accumulated by the thousand. Submitting to all of them is not a growth strategy, it is a way to spend a night producing a spreadsheet.

The parts worth doing are the ones that do not need a stranger's approval: pages that answer a question nobody has answered, data that other people cite, and distribution channels with an API. Those are also the parts that survive the next algorithm update.

If you want the raw material: the dataset is public under CC BY 4.0. Fork it, add your own results, and the acceptance column stops being one person's sample.

FAQ

Are directory submissions worth it at all?

As a one-time seed for a brand-new site, a handful are: the ones with editorial review and real traffic. After that the marginal directory adds a nofollow link on a page with no visitors, and the cost is your attention rather than your money.

Why did only 8 directories accept anything?

Because the other 227 either never received a valid submission, require a human to review an account first, or are aggregators that populate themselves. The dataset marks the reason for each one.

Can I reuse this data?

Yes, CC BY 4.0: cite this page and keep the attribution. The scraper and the field-matching rules are in the same repository.